diff --git a/CMakeLists.txt b/CMakeLists.txt
index 4ddf57ce7f5d..a6e84b86a429 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -185,7 +185,7 @@ assign_source_group("Include" ${GROUP_INCLUDE})
 
 # Source file lists
 file(GLOB_RECURSE COMPILER_SRCS
-    src/auto_schedule/*.cc
+    src/auto_scheduler/*.cc
     src/node/*.cc
     src/ir/*.cc
     src/arith/*.cc
diff --git a/python/tvm/auto_schedule/__init__.py b/python/tvm/auto_scheduler/__init__.py
similarity index 100%
rename from python/tvm/auto_schedule/__init__.py
rename to python/tvm/auto_scheduler/__init__.py
diff --git a/python/tvm/auto_schedule/_ffi_api.py b/python/tvm/auto_scheduler/_ffi_api.py
similarity index 93%
rename from python/tvm/auto_schedule/_ffi_api.py
rename to python/tvm/auto_scheduler/_ffi_api.py
index 9d2b9865ae95..d7b874f71e0f 100644
--- a/python/tvm/auto_schedule/_ffi_api.py
+++ b/python/tvm/auto_scheduler/_ffi_api.py
@@ -15,8 +15,8 @@
 # specific language governing permissions and limitations
 # under the License.
 
-""" Register FFI APIs from C++ for the namespace tvm.auto_schedule. """
+""" Register FFI APIs from C++ for the namespace tvm.auto_scheduler. """
 import tvm._ffi
 
 
-tvm._ffi._init_api("auto_schedule", __name__)
+tvm._ffi._init_api("auto_scheduler", __name__)
diff --git a/python/tvm/auto_schedule/auto_schedule.py b/python/tvm/auto_scheduler/auto_schedule.py
similarity index 90%
rename from python/tvm/auto_schedule/auto_schedule.py
rename to python/tvm/auto_scheduler/auto_schedule.py
index ffbfc3c914ff..d45dbf8d0aaa 100644
--- a/python/tvm/auto_schedule/auto_schedule.py
+++ b/python/tvm/auto_scheduler/auto_schedule.py
@@ -34,7 +34,7 @@
 from . import _ffi_api
 
 
-@tvm._ffi.register_object("auto_schedule.HardwareParams")
+@tvm._ffi.register_object("auto_scheduler.HardwareParams")
 class HardwareParams(Object):
     """ The parameters of target hardware used to guide the search policy
 
@@ -55,7 +55,7 @@ def __init__(self, num_cores, vector_unit_bytes, cache_line_bytes):
                                             vector_unit_bytes, cache_line_bytes)
 
 
-@tvm._ffi.register_object("auto_schedule.SearchTask")
+@tvm._ffi.register_object("auto_scheduler.SearchTask")
 class SearchTask(Object):
     """ The computation information and hardware parameters for a specific schedule search task.
 
@@ -79,12 +79,12 @@ def __init__(self, dag, workload_key, target, target_host=None,
                                             hardware_params)
 
 
-@tvm._ffi.register_object("auto_schedule.SearchPolicy")
+@tvm._ffi.register_object("auto_scheduler.SearchPolicy")
 class SearchPolicy(Object):
     """ The base class of search policies. """
 
 
-@tvm._ffi.register_object("auto_schedule.EmptyPolicy")
+@tvm._ffi.register_object("auto_scheduler.EmptyPolicy")
 class EmptyPolicy(SearchPolicy):
     """ This is an example empty search policy which will always generate
     the init state of ComputeDAG.
@@ -93,7 +93,7 @@ def __init__(self):
         self.__init_handle_by_constructor__(_ffi_api.EmptyPolicy)
 
 
-@tvm._ffi.register_object("auto_schedule.TuningOptions")
+@tvm._ffi.register_object("auto_scheduler.TuningOptions")
 class TuningOptions(Object):
     """ This controls the options of performance tuning.
 
@@ -120,12 +120,12 @@ class TuningOptions(Object):
     measure_callbacks: Optional[List[MeasureCallback]]
       Callback functions called after each measurement.
       Candidates:
-        - auto_schedule.RecordToFile
+        - auto_scheduler.RecordToFile
     pre_search_callbacks: Optional[List[SearchCallback]]
       Callback functions called before the search process.
       Candidates:
-        - auto_schedule.PreloadMeasuredStates
-        - auto_schedule.PreloadCustomSketchRule
+        - auto_scheduler.PreloadMeasuredStates
+        - auto_scheduler.PreloadCustomSketchRule
         TODO(jcf94): Add these implementation in later PRs.
     """
     def __init__(self, num_measure_trials=0, early_stopping=None, num_measures_per_round=64,
@@ -136,7 +136,7 @@ def __init__(self, num_measure_trials=0, early_stopping=None, num_measures_per_r
                 builder = LocalBuilder()
             else:
                 raise ValueError("Invalid builder: " + builder)
-        elif not isinstance(builder, tvm.auto_schedule.measure.ProgramBuilder):
+        elif not isinstance(builder, tvm.auto_scheduler.measure.ProgramBuilder):
             raise ValueError("Invalid builder: " + builder +
                              " . TuningOptions expects a ProgramBuilder or string.")
 
@@ -145,7 +145,7 @@ def __init__(self, num_measure_trials=0, early_stopping=None, num_measures_per_r
                 runner = LocalRunner()
             else:
                 raise ValueError("Invalid runner: " + runner)
-        elif not isinstance(runner, tvm.auto_schedule.measure.ProgramRunner):
+        elif not isinstance(runner, tvm.auto_scheduler.measure.ProgramRunner):
             raise ValueError("Invalid runner: " + runner +
                              " . TuningOptions expects a ProgramRunner or string.")
 
@@ -176,7 +176,7 @@ def auto_schedule(task, search_policy='default', tuning_options=None):
     """
     if not isinstance(task, SearchTask):
         raise ValueError("Invalid task: " + task +
-                         " . `auto_schedule.auto_schedule` expects a SearchTask.")
+                         " . `auto_scheduler.auto_schedule` expects a SearchTask.")
 
     if isinstance(search_policy, str):
         if search_policy == 'default':
@@ -187,7 +187,7 @@ def auto_schedule(task, search_policy='default', tuning_options=None):
             raise ValueError("Invalid search policy: " + search_policy)
     elif not isinstance(search_policy, SearchPolicy):
         raise ValueError("Invalid search policy: " + search_policy +
-                         " . `auto_schedule.auto_schedule` expects a SearchPolicy or a string.")
+                         " . `auto_scheduler.auto_schedule` expects a SearchPolicy or a string.")
 
     sch, tensors = _ffi_api.AutoSchedule(task, search_policy,
                                          tuning_options if tuning_options else TuningOptions())
diff --git a/python/tvm/auto_schedule/compute_dag.py b/python/tvm/auto_scheduler/compute_dag.py
similarity index 99%
rename from python/tvm/auto_schedule/compute_dag.py
rename to python/tvm/auto_scheduler/compute_dag.py
index a4738a933b3e..115d28b4d478 100644
--- a/python/tvm/auto_schedule/compute_dag.py
+++ b/python/tvm/auto_scheduler/compute_dag.py
@@ -30,7 +30,7 @@
 from . import _ffi_api
 
 
-@tvm._ffi.register_object("auto_schedule.ComputeDAG")
+@tvm._ffi.register_object("auto_scheduler.ComputeDAG")
 class ComputeDAG(Object):
     """
     The TVM Auto-scheduler computational graph and related program analyses.
diff --git a/python/tvm/auto_schedule/loop_state.py b/python/tvm/auto_scheduler/loop_state.py
similarity index 97%
rename from python/tvm/auto_schedule/loop_state.py
rename to python/tvm/auto_scheduler/loop_state.py
index 7b8804c8be60..693a668a158e 100644
--- a/python/tvm/auto_schedule/loop_state.py
+++ b/python/tvm/auto_scheduler/loop_state.py
@@ -48,17 +48,17 @@
 from . import _ffi_api
 
 
-@tvm._ffi.register_object("auto_schedule.Iterator")
+@tvm._ffi.register_object("auto_scheduler.Iterator")
 class Iterator(Object):
     """ A loop iterator structure. """
 
 
-@tvm._ffi.register_object("auto_schedule.Stage")
+@tvm._ffi.register_object("auto_scheduler.Stage")
 class Stage(Object):
     """ A stage in the compute declaration. Similar to tvm.te.schedule.Stage. """
 
 
-@tvm._ffi.register_object("auto_schedule.State")
+@tvm._ffi.register_object("auto_scheduler.State")
 class StateObject(Object):
     """ The internal State object """
     def __eq__(self, other):
diff --git a/python/tvm/auto_schedule/measure.py b/python/tvm/auto_scheduler/measure.py
similarity index 96%
rename from python/tvm/auto_schedule/measure.py
rename to python/tvm/auto_scheduler/measure.py
index 24e2af1d8f49..e99c47e6262c 100644
--- a/python/tvm/auto_schedule/measure.py
+++ b/python/tvm/auto_scheduler/measure.py
@@ -54,12 +54,12 @@
 # This can avoid expensive serialization of TVM IR when using multiprocessing.Pool
 GLOBAL_BUILD_ARGUMENTS = None
 
-@tvm._ffi.register_object("auto_schedule.MeasureCallback")
+@tvm._ffi.register_object("auto_scheduler.MeasureCallback")
 class MeasureCallback(Object):
     """ The base class of measurement callback functions. """
 
 
-@tvm._ffi.register_object("auto_schedule.MeasureInput")
+@tvm._ffi.register_object("auto_scheduler.MeasureInput")
 class MeasureInput(Object):
     """ Store the input of a measurement.
 
@@ -74,7 +74,7 @@ def __init__(self, task, state):
         self.__init_handle_by_constructor__(_ffi_api.MeasureInput, task, state.state_object)
 
 
-@tvm._ffi.register_object("auto_schedule.BuildResult")
+@tvm._ffi.register_object("auto_scheduler.BuildResult")
 class BuildResult(Object):
     """ Store the result of a build.
 
@@ -99,7 +99,7 @@ def __init__(self, filename, args, error_no, error_msg, time_cost):
             _ffi_api.BuildResult, filename, args, error_no, error_msg, time_cost)
 
 
-@tvm._ffi.register_object("auto_schedule.MeasureResult")
+@tvm._ffi.register_object("auto_scheduler.MeasureResult")
 class MeasureResult(Object):
     """ Store the results of a measurement.
 
@@ -124,7 +124,7 @@ def __init__(self, costs, error_no, error_msg, all_cost, timestamp):
             error_msg, all_cost, timestamp)
 
 
-@tvm._ffi.register_object("auto_schedule.ProgramBuilder")
+@tvm._ffi.register_object("auto_scheduler.ProgramBuilder")
 class ProgramBuilder(Object):
     """ The base class of ProgramBuilders. """
 
@@ -145,7 +145,7 @@ def build(self, measure_inputs, verbose=1):
         return _ffi_api.ProgramBuilderBuild(self, measure_inputs, verbose)
 
 
-@tvm._ffi.register_object("auto_schedule.ProgramRunner")
+@tvm._ffi.register_object("auto_scheduler.ProgramRunner")
 class ProgramRunner(Object):
     """ The base class of ProgramRunners. """
 
@@ -168,7 +168,7 @@ def run(self, measure_inputs, build_results, verbose=1):
         return _ffi_api.ProgramRunnerRun(self, measure_inputs, build_results, verbose)
 
 
-@tvm._ffi.register_object("auto_schedule.LocalBuilder")
+@tvm._ffi.register_object("auto_scheduler.LocalBuilder")
 class LocalBuilder(ProgramBuilder):
     """ LocalBuilder use local CPU cores to build programs in parallel.
 
@@ -191,7 +191,7 @@ def __init__(self,
             _ffi_api.LocalBuilder, timeout, n_parallel, build_func)
 
 
-@tvm._ffi.register_object("auto_schedule.LocalRunner")
+@tvm._ffi.register_object("auto_scheduler.LocalRunner")
 class LocalRunner(ProgramRunner):
     """ LocalRunner that uses local CPU/GPU to measures the time cost of programs.
 
@@ -334,7 +334,7 @@ def timed_func():
     return res
 
 
-@tvm._ffi.register_func("auto_schedule.local_builder.build")
+@tvm._ffi.register_func("auto_scheduler.local_builder.build")
 def local_builder_build(inputs, timeout, n_parallel, build_func='default', verbose=1):
     """
     Build function of LocalBuilder to build the MeasureInputs to runnable modules.
@@ -376,7 +376,7 @@ def local_builder_build(inputs, timeout, n_parallel, build_func='default', verbo
 
     return results
 
-@tvm._ffi.register_func("auto_schedule.local_runner.run")
+@tvm._ffi.register_func("auto_scheduler.local_runner.run")
 def local_run(inputs, build_results, timeout, number, repeat, min_repeat_ms, cooldown_interval,
               verbose=1):
     """
diff --git a/python/tvm/auto_schedule/measure_record.py b/python/tvm/auto_scheduler/measure_record.py
similarity index 94%
rename from python/tvm/auto_schedule/measure_record.py
rename to python/tvm/auto_scheduler/measure_record.py
index 25a998566280..2a4cd2957a0a 100644
--- a/python/tvm/auto_schedule/measure_record.py
+++ b/python/tvm/auto_scheduler/measure_record.py
@@ -25,7 +25,7 @@
 from . import _ffi_api
 
 
-@tvm._ffi.register_object("auto_schedule.RecordToFile")
+@tvm._ffi.register_object("auto_scheduler.RecordToFile")
 class RecordToFile(MeasureCallback):
     """
     A measurement callback that writes measurement records into a file.
@@ -35,21 +35,21 @@ class RecordToFile(MeasureCallback):
     filename : str
         File name for this callback to write log to.
     """
-    def __init__(self, filename="auto_schedule_tuning.json"):
+    def __init__(self, filename="auto_scheduler_tuning.json"):
         self.__init_handle_by_constructor__(_ffi_api.RecordToFile, filename)
 
 
-@tvm._ffi.register_object("auto_schedule.RecordReader")
+@tvm._ffi.register_object("auto_scheduler.RecordReader")
 class RecordReader(Object):
     """
     Reader of the json log file.
 
     Parameters
     ----------
-    filename : str = "auto_schedule_tuning.json"
+    filename : str = "auto_scheduler_tuning.json"
         File name for this reader to load log from.
     """
-    def __init__(self, filename="auto_schedule_tuning.json"):
+    def __init__(self, filename="auto_scheduler_tuning.json"):
         self.__init_handle_by_constructor__(_ffi_api.RecordReader, filename)
 
     def read_lines(self, max_lines=None, skip_lines=0):
diff --git a/python/tvm/auto_schedule/utils.py b/python/tvm/auto_scheduler/utils.py
similarity index 99%
rename from python/tvm/auto_schedule/utils.py
rename to python/tvm/auto_scheduler/utils.py
index c29675074e22..f7b12028fc36 100644
--- a/python/tvm/auto_schedule/utils.py
+++ b/python/tvm/auto_scheduler/utils.py
@@ -15,7 +15,7 @@
 # specific language governing permissions and limitations
 # under the License.
 
-""" Common utilities for auto_schedule. """
+""" Common utilities for auto_scheduler. """
 
 from typing import Hashable
 import multiprocessing
diff --git a/python/tvm/auto_schedule/workload_registry.py b/python/tvm/auto_scheduler/workload_registry.py
similarity index 97%
rename from python/tvm/auto_schedule/workload_registry.py
rename to python/tvm/auto_scheduler/workload_registry.py
index b50727ec955e..36c203781073 100644
--- a/python/tvm/auto_schedule/workload_registry.py
+++ b/python/tvm/auto_scheduler/workload_registry.py
@@ -55,7 +55,7 @@ def register_workload(func_name, f=None, override=False):
 
     Examples
     --------
-    @auto_schedule.register_workload
+    @auto_scheduler.register_workload
     def matmul(N, M, K):
         A = te.placeholder((N, K), name='A')
         B = te.placeholder((K, M), name='B')
@@ -110,7 +110,7 @@ def make_workload_key(func, args):
 
     if not func_name in WORKLOAD_FUNC_REGISTRY:
         raise ValueError("%s is not registered. "  % func,
-                         "Please register it with @auto_schedule.register_workload")
+                         "Please register it with @auto_scheduler.register_workload")
 
     args = serialize_args(args)
 
@@ -137,11 +137,11 @@ def decode_workload_key_to_func_args(workload_key):
     workload = json.loads(workload_key)
     if not workload[0] in WORKLOAD_FUNC_REGISTRY:
         raise ValueError("%s is not registered. " % workload[0] +
-                         "Please register it with @auto_schedule.register_workload")
+                         "Please register it with @auto_scheduler.register_workload")
     return workload[0], deserialize_args(workload[1:])
 
 
-@tvm._ffi.register_func("auto_schedule.workload_key_to_tensors")
+@tvm._ffi.register_func("auto_scheduler.workload_key_to_tensors")
 def workload_key_to_tensors(workload_key):
     """ Get the input/output tensors from the workload key.
 
diff --git a/src/auto_schedule/auto_schedule.cc b/src/auto_scheduler/auto_schedule.cc
similarity index 88%
rename from src/auto_schedule/auto_schedule.cc
rename to src/auto_scheduler/auto_schedule.cc
index aaf472b1f26a..b515b3accf7a 100644
--- a/src/auto_schedule/auto_schedule.cc
+++ b/src/auto_scheduler/auto_schedule.cc
@@ -18,8 +18,10 @@
  */
 
 /*!
- * \file auto_schedule/auto_schedule.cc
- * \brief The user interface of the TVM Auto-scheduler.
+ * \file auto_scheduler/auto_schedule.cc
+ * \brief The user interface of the TVM Auto-scheduler. This is the entry structure to get
+ * schedule search requirements from upper level (Python API), and returns a high performance
+ * schedule after search process.
  */
 
 #include "auto_schedule.h"
@@ -27,7 +29,7 @@
 #include <tvm/runtime/registry.h>
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 TVM_REGISTER_NODE_TYPE(TuningOptionsNode);
 
@@ -61,7 +63,7 @@ std::pair<te::Schedule, Array<te::Tensor>> AutoSchedule(SearchTask task, SearchP
   return task->compute_dag.ApplySteps(state->transform_steps);
 }
 
-TVM_REGISTER_GLOBAL("auto_schedule.TuningOptions")
+TVM_REGISTER_GLOBAL("auto_scheduler.TuningOptions")
     .set_body_typed([](int num_measure_trials, int early_stopping, int num_measures_per_round,
                        int verbose, ProgramBuilder builder, ProgramRunner runner,
                        Optional<Array<MeasureCallback>> measure_callbacks,
@@ -70,12 +72,12 @@ TVM_REGISTER_GLOBAL("auto_schedule.TuningOptions")
                            builder, runner, measure_callbacks, pre_search_callbacks);
     });
 
-TVM_REGISTER_GLOBAL("auto_schedule.AutoSchedule")
+TVM_REGISTER_GLOBAL("auto_scheduler.AutoSchedule")
     .set_body_typed([](SearchTask task, SearchPolicy search_policy, TuningOptions tuning_options) {
       te::Schedule sch;
       Array<te::Tensor> return_tensors;
       std::tie(sch, return_tensors) = AutoSchedule(task, search_policy, tuning_options);
       return Array<ObjectRef>{sch, return_tensors};
     });
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
diff --git a/src/auto_schedule/auto_schedule.h b/src/auto_scheduler/auto_schedule.h
similarity index 93%
rename from src/auto_schedule/auto_schedule.h
rename to src/auto_scheduler/auto_schedule.h
index d2e49bbe7e4f..55c6992dfd4e 100644
--- a/src/auto_schedule/auto_schedule.h
+++ b/src/auto_scheduler/auto_schedule.h
@@ -18,14 +18,14 @@
  */
 
 /*!
- * \file auto_schedule/auto_schedule.h
+ * \file auto_scheduler/auto_schedule.h
  * \brief The user interface of the TVM Auto-scheduler. This is the entry structure to get
  * schedule search requirements from upper level (Python API), and returns a high performance
  * schedule after search process.
  */
 
-#ifndef TVM_AUTO_SCHEDULE_AUTO_SCHEDULE_H_
-#define TVM_AUTO_SCHEDULE_AUTO_SCHEDULE_H_
+#ifndef TVM_AUTO_SCHEDULER_AUTO_SCHEDULE_H_
+#define TVM_AUTO_SCHEDULER_AUTO_SCHEDULE_H_
 
 #include <utility>
 
@@ -33,7 +33,7 @@
 #include "search_policy/search_policy.h"
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 /*! \brief Tuning and measurement options. */
 class TuningOptionsNode : public Object {
@@ -69,7 +69,7 @@ class TuningOptionsNode : public Object {
     v->Visit("pre_search_callbacks", &pre_search_callbacks);
   }
 
-  static constexpr const char* _type_key = "auto_schedule.TuningOptions";
+  static constexpr const char* _type_key = "auto_scheduler.TuningOptions";
   TVM_DECLARE_FINAL_OBJECT_INFO(TuningOptionsNode, Object);
 };
 
@@ -110,7 +110,7 @@ class TuningOptions : public ObjectRef {
 TVM_DLL std::pair<te::Schedule, Array<te::Tensor>> AutoSchedule(SearchTask task,
                                                                 SearchPolicy search_policy,
                                                                 TuningOptions tuning_options);
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
 
-#endif  // TVM_AUTO_SCHEDULE_AUTO_SCHEDULE_H_
+#endif  // TVM_AUTO_SCHEDULER_AUTO_SCHEDULE_H_
diff --git a/src/auto_schedule/compute_dag.cc b/src/auto_scheduler/compute_dag.cc
similarity index 97%
rename from src/auto_schedule/compute_dag.cc
rename to src/auto_scheduler/compute_dag.cc
index 312a25ad62dd..a7abcb8a7ebf 100644
--- a/src/auto_schedule/compute_dag.cc
+++ b/src/auto_scheduler/compute_dag.cc
@@ -18,7 +18,7 @@
  */
 
 /*!
- * \file auto_schedule/compute_dag.cc
+ * \file auto_scheduler/compute_dag.cc
  * \brief Compute declaration graph and its related analysis tools.
  */
 
@@ -40,7 +40,7 @@
 #include "utils.h"
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 using namespace tvm::tir;
 
@@ -450,11 +450,11 @@ TVM_STATIC_IR_FUNCTOR(ReprPrinter, vtable)
       p->stream << ss.str();
     });
 
-TVM_REGISTER_GLOBAL("auto_schedule.ComputeDAG").set_body_typed([](Array<te::Tensor> tensors) {
+TVM_REGISTER_GLOBAL("auto_scheduler.ComputeDAG").set_body_typed([](Array<te::Tensor> tensors) {
   return ComputeDAG(tensors);
 });
 
-TVM_REGISTER_GLOBAL("auto_schedule.ComputeDAGApplyStepsFromState")
+TVM_REGISTER_GLOBAL("auto_scheduler.ComputeDAGApplyStepsFromState")
     .set_body_typed([](const ComputeDAG& dag, const State& state) {
       te::Schedule sch;
       Array<te::Tensor> return_tensors;
@@ -462,15 +462,15 @@ TVM_REGISTER_GLOBAL("auto_schedule.ComputeDAGApplyStepsFromState")
       return Array<ObjectRef>{sch, return_tensors};
     });
 
-TVM_REGISTER_GLOBAL("auto_schedule.ComputeDAGPrintPythonCodeFromState")
+TVM_REGISTER_GLOBAL("auto_scheduler.ComputeDAGPrintPythonCodeFromState")
     .set_body_typed([](const ComputeDAG& dag, const State& state) {
       return dag.PrintStepsAsPython(state->transform_steps);
     });
 
-TVM_REGISTER_GLOBAL("auto_schedule.ComputeDAGInferBoundFromState")
+TVM_REGISTER_GLOBAL("auto_scheduler.ComputeDAGInferBoundFromState")
     .set_body_typed([](const ComputeDAG& dag, const State& state) {
       return dag.InferBound(state);
     });
 
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
diff --git a/src/auto_schedule/compute_dag.h b/src/auto_scheduler/compute_dag.h
similarity index 93%
rename from src/auto_schedule/compute_dag.h
rename to src/auto_scheduler/compute_dag.h
index bb582d32ee7e..2417d72983b0 100644
--- a/src/auto_schedule/compute_dag.h
+++ b/src/auto_scheduler/compute_dag.h
@@ -18,7 +18,7 @@
  */
 
 /*!
- * \file auto_schedule/compute_dag.h
+ * \file auto_scheduler/compute_dag.h
  * \brief The TVM Auto-scheduler computational graph and related program analyses.
  *
  * We convert a compute declaration described by `tvm.compute` (could be a single operator or a
@@ -32,8 +32,8 @@
  * `LoopState` with extra information got from TVM schedule ...).
  */
 
-#ifndef TVM_AUTO_SCHEDULE_COMPUTE_DAG_H_
-#define TVM_AUTO_SCHEDULE_COMPUTE_DAG_H_
+#ifndef TVM_AUTO_SCHEDULER_COMPUTE_DAG_H_
+#define TVM_AUTO_SCHEDULER_COMPUTE_DAG_H_
 
 #include <tvm/te/schedule.h>
 
@@ -42,7 +42,7 @@
 #include "loop_state.h"
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 /*! \brief The TVM Auto-scheduler computational graph and related program analyses. */
 class ComputeDAGNode : public Object {
@@ -67,7 +67,7 @@ class ComputeDAGNode : public Object {
     v->Visit("init_state", &init_state);
   }
 
-  static constexpr const char* _type_key = "auto_schedule.ComputeDAG";
+  static constexpr const char* _type_key = "auto_scheduler.ComputeDAG";
   TVM_DECLARE_FINAL_OBJECT_INFO(ComputeDAGNode, Object);
 };
 
@@ -118,7 +118,7 @@ class ComputeDAG : public ObjectRef {
   TVM_DEFINE_OBJECT_REF_COW_METHOD(ComputeDAGNode);
 };
 
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
 
-#endif  // TVM_AUTO_SCHEDULE_COMPUTE_DAG_H_
+#endif  // TVM_AUTO_SCHEDULER_COMPUTE_DAG_H_
diff --git a/src/auto_schedule/loop_state.cc b/src/auto_scheduler/loop_state.cc
similarity index 97%
rename from src/auto_schedule/loop_state.cc
rename to src/auto_scheduler/loop_state.cc
index 666efcebce04..1bfcb9ebc58a 100644
--- a/src/auto_schedule/loop_state.cc
+++ b/src/auto_scheduler/loop_state.cc
@@ -18,9 +18,9 @@
  */
 
 /*!
- * \file auto_schedule/loop_state.cc
+ * \file auto_scheduler/loop_state.cc
  * \brief An lightweight IR (intermediate representation) for loop structures.
- * see auto_schedule/loop_state.h for more explanation.
+ * see auto_scheduler/loop_state.h for more explanation.
  */
 
 #include "loop_state.h"
@@ -34,7 +34,7 @@
 #include "utils.h"
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 TVM_REGISTER_OBJECT_TYPE(StepNode);
 TVM_REGISTER_NODE_TYPE(StageNode);
@@ -386,28 +386,28 @@ TVM_STATIC_IR_FUNCTOR(ReprPrinter, vtable)
     });
 
 /********** State interface API for ffi **********/
-TVM_REGISTER_GLOBAL("auto_schedule.StateReorder")
+TVM_REGISTER_GLOBAL("auto_scheduler.StateReorder")
     .set_body_typed([](State state, int stage_id, const Array<Iterator>& order) {
       state.reorder(stage_id, order);
       return state;
     });
 
-TVM_REGISTER_GLOBAL("auto_schedule.StateSplit")
+TVM_REGISTER_GLOBAL("auto_scheduler.StateSplit")
     .set_body_typed([](State state, int stage_id, const Iterator& it,
                        const Array<Optional<Integer>>& lengths, bool inner_to_outer) {
       const auto& res = state.split(stage_id, it, lengths, inner_to_outer);
       return Array<ObjectRef>{state, res};
     });
 
-TVM_REGISTER_GLOBAL("auto_schedule.StateFuse")
+TVM_REGISTER_GLOBAL("auto_scheduler.StateFuse")
     .set_body_typed([](State state, int stage_id, const Array<Iterator>& iters) {
       const auto& res = state.fuse(stage_id, iters);
       return Array<ObjectRef>{state, res};
     });
 
-TVM_REGISTER_GLOBAL("auto_schedule.StateEqual").set_body_typed([](State state1, State state2) {
+TVM_REGISTER_GLOBAL("auto_scheduler.StateEqual").set_body_typed([](State state1, State state2) {
   return std::equal_to<State>()(state1, state2);
 });
 
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
diff --git a/src/auto_schedule/loop_state.h b/src/auto_scheduler/loop_state.h
similarity index 93%
rename from src/auto_schedule/loop_state.h
rename to src/auto_scheduler/loop_state.h
index 5ba47b7263a1..04e5304b6943 100644
--- a/src/auto_schedule/loop_state.h
+++ b/src/auto_scheduler/loop_state.h
@@ -18,7 +18,7 @@
  */
 
 /*!
- * \file auto_schedule/loop_state.h
+ * \file auto_scheduler/loop_state.h
  * \brief The definition of the "state" in search.
  *
  * Each LoopState corresponds to a schedule for its ComputeDAG.
@@ -45,8 +45,8 @@
  * copy on write style. All objects are immutable, which is similar to TVM IR.
  */
 
-#ifndef TVM_AUTO_SCHEDULE_LOOP_STATE_H_
-#define TVM_AUTO_SCHEDULE_LOOP_STATE_H_
+#ifndef TVM_AUTO_SCHEDULER_LOOP_STATE_H_
+#define TVM_AUTO_SCHEDULER_LOOP_STATE_H_
 
 #include <tvm/runtime/container.h>
 
@@ -55,7 +55,7 @@
 #include "transform_step.h"
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 using namespace tvm::tir;
 
@@ -135,7 +135,7 @@ class IteratorNode : public Object {
     v->Visit("range", &range);
   }
 
-  static constexpr const char* _type_key = "auto_schedule.Iterator";
+  static constexpr const char* _type_key = "auto_scheduler.Iterator";
   TVM_DECLARE_FINAL_OBJECT_INFO(IteratorNode, Object);
 };
 
@@ -187,7 +187,7 @@ class StageNode : public Object {
     v->Visit("iters", &iters);
   }
 
-  static constexpr const char* _type_key = "auto_schedule.Stage";
+  static constexpr const char* _type_key = "auto_scheduler.Stage";
   TVM_DECLARE_FINAL_OBJECT_INFO(StageNode, Object);
 };
 
@@ -241,7 +241,7 @@ class StateNode : public Object {
     v->Visit("concrete", &concrete);
   }
 
-  static constexpr const char* _type_key = "auto_schedule.State";
+  static constexpr const char* _type_key = "auto_scheduler.State";
   TVM_DECLARE_FINAL_OBJECT_INFO(StateNode, Object);
 
  private:
@@ -347,22 +347,22 @@ class State : public ObjectRef {
                                     const Array<Optional<Integer>>& lengths, bool inner_to_outer);
 };
 
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
 
 // Hash and equal function for State
 namespace std {
 
-/*! \brief The hash function for auto_schedule::State. */
+/*! \brief The hash function for auto_scheduler::State. */
 template <>
-struct hash<::tvm::auto_schedule::State> {
-  std::size_t operator()(const ::tvm::auto_schedule::State& state) const {
+struct hash<::tvm::auto_scheduler::State> {
+  std::size_t operator()(const ::tvm::auto_scheduler::State& state) const {
     return tvm::runtime::ObjectHash()(state.ToStr());
   }
 };
 
 /*!
- * \brief The equal_to function for auto_schedule::State.
+ * \brief The equal_to function for auto_scheduler::State.
  * We use the schedule result(its string format) of a state to check if two states are `euqal`.
  * Equal States: 1. the transform steps are totally the same; 2. even with different steps, two
  * states may still result in a same schedule. e.g. To split a axis with extent 512 to 3 parts
@@ -370,13 +370,13 @@ struct hash<::tvm::auto_schedule::State> {
  * to split from outter to inner by factors [8, 16])
  */
 template <>
-struct equal_to<::tvm::auto_schedule::State> {
-  bool operator()(const ::tvm::auto_schedule::State& lhs,
-                  const ::tvm::auto_schedule::State& rhs) const {
+struct equal_to<::tvm::auto_scheduler::State> {
+  bool operator()(const ::tvm::auto_scheduler::State& lhs,
+                  const ::tvm::auto_scheduler::State& rhs) const {
     return lhs.ToStr() == rhs.ToStr();
   }
 };
 
 }  // namespace std
 
-#endif  // TVM_AUTO_SCHEDULE_LOOP_STATE_H_
+#endif  // TVM_AUTO_SCHEDULER_LOOP_STATE_H_
diff --git a/src/auto_schedule/measure.cc b/src/auto_scheduler/measure.cc
similarity index 93%
rename from src/auto_schedule/measure.cc
rename to src/auto_scheduler/measure.cc
index b710745b02f9..3c545525f4e6 100644
--- a/src/auto_schedule/measure.cc
+++ b/src/auto_scheduler/measure.cc
@@ -18,7 +18,7 @@
  */
 
 /*!
- * \file auto_schedule/measure.cc
+ * \file auto_scheduler/measure.cc
  * \brief Distributed measurement infrastructure to measure the runtime costs of tensor programs.
  */
 
@@ -31,7 +31,7 @@
 #include "utils.h"
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 TVM_REGISTER_NODE_TYPE(MeasureInputNode);
 TVM_REGISTER_NODE_TYPE(BuildResultNode);
@@ -111,11 +111,11 @@ LocalBuilder::LocalBuilder(int timeout, int n_parallel, const String& build_func
 }
 
 Array<BuildResult> LocalBuilderNode::Build(const Array<MeasureInput>& inputs, int verbose) {
-  if (const auto* f = runtime::Registry::Get("auto_schedule.local_builder.build")) {
+  if (const auto* f = runtime::Registry::Get("auto_scheduler.local_builder.build")) {
     Array<BuildResult> results = (*f)(inputs, timeout, n_parallel, build_func, verbose);
     return results;
   }
-  LOG(FATAL) << "auto_schedule.local_builder.build is not registered. "
+  LOG(FATAL) << "auto_scheduler.local_builder.build is not registered. "
              << "This is a function registered in Python, "
              << "make sure the TVM Python runtime has been loaded successfully.";
   throw;
@@ -135,12 +135,12 @@ LocalRunner::LocalRunner(int timeout, int number, int repeat, int min_repeat_ms,
 
 Array<MeasureResult> LocalRunnerNode::Run(const Array<MeasureInput>& inputs,
                                           const Array<BuildResult>& build_results, int verbose) {
-  if (const auto* f = runtime::Registry::Get("auto_schedule.local_runner.run")) {
+  if (const auto* f = runtime::Registry::Get("auto_scheduler.local_runner.run")) {
     Array<MeasureResult> results = (*f)(inputs, build_results, timeout, number, repeat,
                                         min_repeat_ms, cooldown_interval, verbose);
     return results;
   }
-  LOG(FATAL) << "auto_schedule.local_runner.run is not registered. "
+  LOG(FATAL) << "auto_scheduler.local_runner.run is not registered. "
              << "This is a function registered in Python, "
              << "make sure the TVM Python runtime has been loaded successfully.";
   throw;
@@ -291,41 +291,41 @@ TVM_STATIC_IR_FUNCTOR(ReprPrinter, vtable)
     });
 
 /********** Measure interface API for ffi **********/
-TVM_REGISTER_GLOBAL("auto_schedule.MeasureInput").set_body_typed([](SearchTask task, State state) {
+TVM_REGISTER_GLOBAL("auto_scheduler.MeasureInput").set_body_typed([](SearchTask task, State state) {
   return MeasureInput(task, state);
 });
 
-TVM_REGISTER_GLOBAL("auto_schedule.BuildResult")
+TVM_REGISTER_GLOBAL("auto_scheduler.BuildResult")
     .set_body_typed([](String filename, Array<te::Tensor> args, int error_no, String error_msg,
                        double time_cost) {
       return BuildResult(filename, args, error_no, error_msg, time_cost);
     });
 
-TVM_REGISTER_GLOBAL("auto_schedule.MeasureResult")
+TVM_REGISTER_GLOBAL("auto_scheduler.MeasureResult")
     .set_body_typed([](Array<PrimExpr> costs, int error_no, String error_msg, double all_cost,
                        double timestamp) {
       return MeasureResult(costs, error_no, error_msg, all_cost, timestamp);
     });
 
-TVM_REGISTER_GLOBAL("auto_schedule.ProgramBuilderBuild")
+TVM_REGISTER_GLOBAL("auto_scheduler.ProgramBuilderBuild")
     .set_body_typed([](const ProgramBuilder& builder, const Array<MeasureInput>& inputs,
                        int verbose) { return builder->Build(inputs, verbose); });
 
-TVM_REGISTER_GLOBAL("auto_schedule.ProgramRunnerRun")
+TVM_REGISTER_GLOBAL("auto_scheduler.ProgramRunnerRun")
     .set_body_typed([](const ProgramRunner& runner, const Array<MeasureInput>& inputs,
                        const Array<BuildResult>& build_results,
                        int verbose) { return runner->Run(inputs, build_results, verbose); });
 
-TVM_REGISTER_GLOBAL("auto_schedule.LocalBuilder")
+TVM_REGISTER_GLOBAL("auto_scheduler.LocalBuilder")
     .set_body_typed([](int timeout, int n_parallel, const String& build_func) {
       return LocalBuilder(timeout, n_parallel, build_func);
     });
 
-TVM_REGISTER_GLOBAL("auto_schedule.LocalRunner")
+TVM_REGISTER_GLOBAL("auto_scheduler.LocalRunner")
     .set_body_typed([](int timeout, int number, int repeat, int min_repeat_ms,
                        double cooldown_interval) {
       return LocalRunner(timeout, number, repeat, min_repeat_ms, cooldown_interval);
     });
 
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
diff --git a/src/auto_schedule/measure.h b/src/auto_scheduler/measure.h
similarity index 94%
rename from src/auto_schedule/measure.h
rename to src/auto_scheduler/measure.h
index a7890eaffd0d..50b6fcfd6520 100644
--- a/src/auto_schedule/measure.h
+++ b/src/auto_scheduler/measure.h
@@ -18,7 +18,7 @@
  */
 
 /*!
- * \file auto_schedule/measure.h
+ * \file auto_scheduler/measure.h
  * \brief Distributed measurement infrastructure to measure the runtime costs of tensor programs.
  * These functions are responsible for building the tvm module, uploading it to remote devices,
  * recording the running time costs, and checking the correctness of the output.
@@ -33,8 +33,8 @@
  * We implement these in python to utilize python's multiprocessing and error handling.
  */
 
-#ifndef TVM_AUTO_SCHEDULE_MEASURE_H_
-#define TVM_AUTO_SCHEDULE_MEASURE_H_
+#ifndef TVM_AUTO_SCHEDULER_MEASURE_H_
+#define TVM_AUTO_SCHEDULER_MEASURE_H_
 
 #include <string>
 #include <unordered_map>
@@ -44,7 +44,7 @@
 #include "search_task.h"
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 class SearchPolicy;
 class MeasureInput;
@@ -90,7 +90,7 @@ class MeasureInputNode : public Object {
   /*! \brief Do shallow copy. */
   MeasureInput copy() const;
 
-  static constexpr const char* _type_key = "auto_schedule.MeasureInput";
+  static constexpr const char* _type_key = "auto_scheduler.MeasureInput";
   TVM_DECLARE_FINAL_OBJECT_INFO(MeasureInputNode, Object);
 };
 
@@ -132,7 +132,7 @@ class BuildResultNode : public Object {
     v->Visit("time_cost", &time_cost);
   }
 
-  static constexpr const char* _type_key = "auto_schedule.BuildResult";
+  static constexpr const char* _type_key = "auto_scheduler.BuildResult";
   TVM_DECLARE_FINAL_OBJECT_INFO(BuildResultNode, Object);
 };
 
@@ -180,7 +180,7 @@ class MeasureResultNode : public Object {
   /*! \brief Do shallow copy. */
   MeasureResult copy() const;
 
-  static constexpr const char* _type_key = "auto_schedule.MeasureResult";
+  static constexpr const char* _type_key = "auto_scheduler.MeasureResult";
   TVM_DECLARE_FINAL_OBJECT_INFO(MeasureResultNode, Object);
 };
 
@@ -216,7 +216,7 @@ class MeasureCallbackNode : public Object {
    */
   virtual void Callback(const SearchPolicy& policy, const Array<MeasureInput>& inputs,
                         const Array<MeasureResult>& results) = 0;
-  static constexpr const char* _type_key = "auto_schedule.MeasureCallback";
+  static constexpr const char* _type_key = "auto_scheduler.MeasureCallback";
   TVM_DECLARE_BASE_OBJECT_INFO(MeasureCallbackNode, Object);
 };
 
@@ -248,7 +248,7 @@ class ProgramBuilderNode : public Object {
    */
   virtual Array<BuildResult> Build(const Array<MeasureInput>& inputs, int verbose) = 0;
 
-  static constexpr const char* _type_key = "auto_schedule.ProgramBuilder";
+  static constexpr const char* _type_key = "auto_scheduler.ProgramBuilder";
   TVM_DECLARE_BASE_OBJECT_INFO(ProgramBuilderNode, Object);
 };
 
@@ -278,7 +278,7 @@ class ProgramRunnerNode : public Object {
   virtual Array<MeasureResult> Run(const Array<MeasureInput>& inputs,
                                    const Array<BuildResult>& build_results, int verbose) = 0;
 
-  static constexpr const char* _type_key = "auto_schedule.ProgramRunner";
+  static constexpr const char* _type_key = "auto_scheduler.ProgramRunner";
   TVM_DECLARE_BASE_OBJECT_INFO(ProgramRunnerNode, Object);
 };
 
@@ -301,7 +301,7 @@ class LocalBuilderNode : public ProgramBuilderNode {
 
   Array<BuildResult> Build(const Array<MeasureInput>& inputs, int verbose) final;
 
-  static constexpr const char* _type_key = "auto_schedule.LocalBuilder";
+  static constexpr const char* _type_key = "auto_scheduler.LocalBuilder";
   TVM_DECLARE_FINAL_OBJECT_INFO(LocalBuilderNode, ProgramBuilderNode);
 };
 
@@ -338,7 +338,7 @@ class LocalRunnerNode : public ProgramRunnerNode {
   Array<MeasureResult> Run(const Array<MeasureInput>& inputs,
                            const Array<BuildResult>& build_results, int verbose) final;
 
-  static constexpr const char* _type_key = "auto_schedule.LocalRunner";
+  static constexpr const char* _type_key = "auto_scheduler.LocalRunner";
   TVM_DECLARE_FINAL_OBJECT_INFO(LocalRunnerNode, ProgramRunnerNode);
 };
 
@@ -349,7 +349,7 @@ class LocalRunnerNode : public ProgramRunnerNode {
 class LocalRunner : public ProgramRunner {
  public:
   /*!
-   * \brief The constructor. See the corresponding class in python/tvm/auto_schedule/measure.py
+   * \brief The constructor. See the corresponding class in python/tvm/auto_scheduler/measure.py
    * for more detailed parameter explaination.
    * \param timeout The timeout limit (in second) for each run.
    * This is used in a wrapper of the multiprocessing.Process.join().
@@ -416,7 +416,7 @@ class ProgramMeasurerNode : public Object {
   /*! \brief The default max continuous error setting. */
   static const int DEFAULT_MAX_CONTINOUS_ERROR = 150;
 
-  static constexpr const char* _type_key = "auto_schedule.ProgramMeasurer";
+  static constexpr const char* _type_key = "auto_scheduler.ProgramMeasurer";
   TVM_DECLARE_FINAL_OBJECT_INFO(ProgramMeasurerNode, Object);
 };
 
@@ -442,7 +442,7 @@ class ProgramMeasurer : public ObjectRef {
   TVM_DEFINE_MUTABLE_OBJECT_REF_METHODS(ProgramMeasurer, ObjectRef, ProgramMeasurerNode);
 };
 
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
 
-#endif  // TVM_AUTO_SCHEDULE_MEASURE_H_
+#endif  // TVM_AUTO_SCHEDULER_MEASURE_H_
diff --git a/src/auto_schedule/measure_record.cc b/src/auto_scheduler/measure_record.cc
similarity index 82%
rename from src/auto_schedule/measure_record.cc
rename to src/auto_scheduler/measure_record.cc
index 99bd5917f7c8..f6f882edb5a2 100644
--- a/src/auto_schedule/measure_record.cc
+++ b/src/auto_scheduler/measure_record.cc
@@ -18,7 +18,7 @@
  */
 
 /*!
- * \file auto_schedule/measure_record.cc
+ * \file auto_scheduler/measure_record.cc
  * \brief Json serialization format for dumping and loading tuning records.
  */
 
@@ -62,14 +62,14 @@ inline std::vector<int> IntArrayToVector(
 }
 
 template <>
-struct Handler<::tvm::Array<::tvm::auto_schedule::Stage>> {
+struct Handler<::tvm::Array<::tvm::auto_scheduler::Stage>> {
   inline static void Write(dmlc::JSONWriter* writer,
-                           const ::tvm::Array<::tvm::auto_schedule::Stage>& data) {
+                           const ::tvm::Array<::tvm::auto_scheduler::Stage>& data) {
     writer->BeginArray(false);
     writer->EndArray();
   }
   inline static void Read(dmlc::JSONReader* reader,
-                          ::tvm::Array<::tvm::auto_schedule::Stage>* data) {
+                          ::tvm::Array<::tvm::auto_scheduler::Stage>* data) {
     bool s;
     reader->BeginArray();
     s = reader->NextArrayItem();
@@ -78,26 +78,26 @@ struct Handler<::tvm::Array<::tvm::auto_schedule::Stage>> {
 };
 
 template <>
-struct Handler<::tvm::Array<::tvm::auto_schedule::Step>> {
+struct Handler<::tvm::Array<::tvm::auto_scheduler::Step>> {
   inline static void Write(dmlc::JSONWriter* writer,
-                           const ::tvm::Array<::tvm::auto_schedule::Step>& data) {
+                           const ::tvm::Array<::tvm::auto_scheduler::Step>& data) {
     writer->BeginArray(false);
     for (size_t i = 0; i < data.size(); ++i) {
       writer->WriteArraySeperator();
       writer->BeginArray(false);
-      if (auto ps = data[i].as<::tvm::auto_schedule::ReorderStepNode>()) {
+      if (auto ps = data[i].as<::tvm::auto_scheduler::ReorderStepNode>()) {
         writer->WriteArrayItem(std::string("RE"));
         writer->WriteArrayItem(ps->stage_id);
         writer->WriteArrayItem(IntArrayToVector(ps->after_ids));
-      } else if (auto ps = data[i].as<::tvm::auto_schedule::SplitStepNode>()) {
+      } else if (auto ps = data[i].as<::tvm::auto_scheduler::SplitStepNode>()) {
         writer->WriteArrayItem(std::string("SP"));
         writer->WriteArrayItem(ps->stage_id);
         writer->WriteArrayItem(ps->iter_id);
-        writer->WriteArrayItem(ps->extent ? ::tvm::auto_schedule::GetIntImm(ps->extent.value())
+        writer->WriteArrayItem(ps->extent ? ::tvm::auto_scheduler::GetIntImm(ps->extent.value())
                                           : 0);
         writer->WriteArrayItem(IntArrayToVector(ps->lengths));
         writer->WriteArrayItem(static_cast<int>(ps->inner_to_outer));
-      } else if (auto ps = data[i].as<::tvm::auto_schedule::FuseStepNode>()) {
+      } else if (auto ps = data[i].as<::tvm::auto_scheduler::FuseStepNode>()) {
         writer->WriteArrayItem(std::string("FU"));
         writer->WriteArrayItem(ps->stage_id);
         writer->WriteArrayItem(IntArrayToVector(ps->fused_ids));
@@ -110,7 +110,7 @@ struct Handler<::tvm::Array<::tvm::auto_schedule::Step>> {
   }
 
   inline static void Read(dmlc::JSONReader* reader,
-                          ::tvm::Array<::tvm::auto_schedule::Step>* data) {
+                          ::tvm::Array<::tvm::auto_scheduler::Step>* data) {
     std::vector<int> int_list;
     bool s, inner_to_outer;
     std::string name, scope_name, pragma_type, ti_func_name;
@@ -134,7 +134,7 @@ struct Handler<::tvm::Array<::tvm::auto_schedule::Step>> {
         for (const auto& i : int_list) {
           after_ids.push_back(i);
         }
-        data->push_back(::tvm::auto_schedule::ReorderStep(stage_id, after_ids));
+        data->push_back(::tvm::auto_scheduler::ReorderStep(stage_id, after_ids));
       } else if (name == "SP") {
         s = reader->NextArrayItem();
         CHECK(s);
@@ -155,7 +155,7 @@ struct Handler<::tvm::Array<::tvm::auto_schedule::Step>> {
         for (const auto& i : int_list) {
           lengths.push_back(::tvm::Integer(i));
         }
-        data->push_back(::tvm::auto_schedule::SplitStep(
+        data->push_back(::tvm::auto_scheduler::SplitStep(
             stage_id, iter_id, extent == 0 ? ::tvm::PrimExpr() : extent, lengths, inner_to_outer));
       } else if (name == "FU") {
         s = reader->NextArrayItem();
@@ -168,7 +168,7 @@ struct Handler<::tvm::Array<::tvm::auto_schedule::Step>> {
         for (const auto& i : int_list) {
           fused_ids.push_back(i);
         }
-        data->push_back(::tvm::auto_schedule::FuseStep(stage_id, fused_ids));
+        data->push_back(::tvm::auto_scheduler::FuseStep(stage_id, fused_ids));
       } else {
         LOG(FATAL) << "Invalid step format";
       }
@@ -179,14 +179,14 @@ struct Handler<::tvm::Array<::tvm::auto_schedule::Step>> {
 };
 
 template <>
-struct Handler<::tvm::auto_schedule::StateNode> {
-  inline static void Write(dmlc::JSONWriter* writer, const ::tvm::auto_schedule::StateNode& data) {
+struct Handler<::tvm::auto_scheduler::StateNode> {
+  inline static void Write(dmlc::JSONWriter* writer, const ::tvm::auto_scheduler::StateNode& data) {
     writer->BeginArray(false);
     writer->WriteArrayItem(data.stages);
     writer->WriteArrayItem(data.transform_steps);
     writer->EndArray();
   }
-  inline static void Read(dmlc::JSONReader* reader, ::tvm::auto_schedule::StateNode* data) {
+  inline static void Read(dmlc::JSONReader* reader, ::tvm::auto_scheduler::StateNode* data) {
     reader->BeginArray();
     bool s;
     s = reader->NextArrayItem();
@@ -201,15 +201,15 @@ struct Handler<::tvm::auto_schedule::StateNode> {
 };
 
 template <>
-struct Handler<::tvm::auto_schedule::SearchTaskNode> {
+struct Handler<::tvm::auto_scheduler::SearchTaskNode> {
   inline static void Write(dmlc::JSONWriter* writer,
-                           const ::tvm::auto_schedule::SearchTaskNode& data) {
+                           const ::tvm::auto_scheduler::SearchTaskNode& data) {
     writer->BeginArray(false);
     writer->WriteArrayItem(std::string(data.workload_key));
     writer->WriteArrayItem(data.target->str());
     writer->EndArray();
   }
-  inline static void Read(dmlc::JSONReader* reader, ::tvm::auto_schedule::SearchTaskNode* data) {
+  inline static void Read(dmlc::JSONReader* reader, ::tvm::auto_scheduler::SearchTaskNode* data) {
     std::string target_str;
     bool s;
 
@@ -228,18 +228,18 @@ struct Handler<::tvm::auto_schedule::SearchTaskNode> {
 };
 
 template <>
-struct Handler<::tvm::auto_schedule::MeasureInputNode> {
+struct Handler<::tvm::auto_scheduler::MeasureInputNode> {
   inline static void Write(dmlc::JSONWriter* writer,
-                           const ::tvm::auto_schedule::MeasureInputNode& data) {
+                           const ::tvm::auto_scheduler::MeasureInputNode& data) {
     writer->BeginArray(false);
     writer->WriteArrayItem(*data.task.operator->());
     writer->WriteArrayItem(*data.state.operator->());
     writer->EndArray();
   }
-  inline static void Read(dmlc::JSONReader* reader, ::tvm::auto_schedule::MeasureInputNode* data) {
+  inline static void Read(dmlc::JSONReader* reader, ::tvm::auto_scheduler::MeasureInputNode* data) {
     bool s;
-    auto task_node = ::tvm::make_object<::tvm::auto_schedule::SearchTaskNode>();
-    auto state_node = ::tvm::make_object<::tvm::auto_schedule::StateNode>();
+    auto task_node = ::tvm::make_object<::tvm::auto_scheduler::SearchTaskNode>();
+    auto state_node = ::tvm::make_object<::tvm::auto_scheduler::StateNode>();
     state_node->concrete = true;
 
     reader->BeginArray();
@@ -252,15 +252,15 @@ struct Handler<::tvm::auto_schedule::MeasureInputNode> {
     s = reader->NextArrayItem();
     CHECK(!s);
 
-    data->task = ::tvm::auto_schedule::SearchTask(task_node);
-    data->state = ::tvm::auto_schedule::State(state_node);
+    data->task = ::tvm::auto_scheduler::SearchTask(task_node);
+    data->state = ::tvm::auto_scheduler::State(state_node);
   }
 };
 
 template <>
-struct Handler<::tvm::auto_schedule::MeasureResultNode> {
+struct Handler<::tvm::auto_scheduler::MeasureResultNode> {
   inline static void Write(dmlc::JSONWriter* writer,
-                           const ::tvm::auto_schedule::MeasureResultNode& data) {
+                           const ::tvm::auto_scheduler::MeasureResultNode& data) {
     writer->BeginArray(false);
     writer->WriteArraySeperator();
     writer->BeginArray(false);
@@ -275,7 +275,8 @@ struct Handler<::tvm::auto_schedule::MeasureResultNode> {
     writer->WriteArrayItem(static_cast<int>((data.timestamp)));
     writer->EndArray();
   }
-  inline static void Read(dmlc::JSONReader* reader, ::tvm::auto_schedule::MeasureResultNode* data) {
+  inline static void Read(dmlc::JSONReader* reader,
+                          ::tvm::auto_scheduler::MeasureResultNode* data) {
     bool s;
     std::vector<double> tmp;
 
@@ -305,12 +306,12 @@ struct Handler<::tvm::auto_schedule::MeasureResultNode> {
 }  // namespace dmlc
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 TVM_REGISTER_OBJECT_TYPE(RecordToFileNode);
 TVM_REGISTER_OBJECT_TYPE(RecordReaderNode);
 
-const std::string AUTO_SCHEDULE_LOG_VERSION = "v0.2";  // NOLINT(*)
+const std::string AUTO_SCHEDULER_LOG_VERSION = "v0.2";  // NOLINT(*)
 
 RecordToFile::RecordToFile(String filename) {
   auto node = make_object<RecordToFileNode>();
@@ -325,7 +326,7 @@ void WriteMeasureRecords(std::ostream* os, const Array<MeasureInput>& inputs,
     writer.BeginObject(false);
     writer.WriteObjectKeyValue("i", *inputs[i].operator->());
     writer.WriteObjectKeyValue("r", *results[i].operator->());
-    writer.WriteObjectKeyValue("v", AUTO_SCHEDULE_LOG_VERSION);
+    writer.WriteObjectKeyValue("v", AUTO_SCHEDULER_LOG_VERSION);
     writer.EndObject();
     *os << "\n";
   }
@@ -405,21 +406,21 @@ std::pair<Array<MeasureInput>, Array<MeasureResult>> RecordReaderNode::ReadLines
   return std::make_pair(inputs, results);
 }
 
-TVM_REGISTER_GLOBAL("auto_schedule.RecordToFile").set_body_typed([](const String& filename) {
+TVM_REGISTER_GLOBAL("auto_scheduler.RecordToFile").set_body_typed([](const String& filename) {
   return RecordToFile(filename);
 });
 
-TVM_REGISTER_GLOBAL("auto_schedule.RecordReader").set_body_typed([](const String& filename) {
+TVM_REGISTER_GLOBAL("auto_scheduler.RecordReader").set_body_typed([](const String& filename) {
   return RecordReader(filename);
 });
 
-TVM_REGISTER_GLOBAL("auto_schedule.RecordReaderReadLines")
+TVM_REGISTER_GLOBAL("auto_scheduler.RecordReaderReadLines")
     .set_body_typed([](RecordReader reader, int size, int skip_size) {
       const auto& res = reader->ReadLines(size, skip_size);
       return Array<ObjectRef>{res.first, res.second};
     });
 
-TVM_REGISTER_GLOBAL("auto_schedule.RecordReaderReadNext").set_body_typed([](RecordReader reader) {
+TVM_REGISTER_GLOBAL("auto_scheduler.RecordReaderReadNext").set_body_typed([](RecordReader reader) {
   auto inp = make_object<MeasureInputNode>();
   auto res = make_object<MeasureResultNode>();
   if (reader->ReadNext(inp.get(), res.get())) {
@@ -429,10 +430,10 @@ TVM_REGISTER_GLOBAL("auto_schedule.RecordReaderReadNext").set_body_typed([](Reco
   }
 });
 
-TVM_REGISTER_GLOBAL("auto_schedule.SaveRecords")
+TVM_REGISTER_GLOBAL("auto_scheduler.SaveRecords")
     .set_body_typed([](String filename, Array<MeasureInput> in, Array<MeasureResult> res) {
       std::ofstream ofs(filename, std::ofstream::app);
       WriteMeasureRecords(&ofs, in, res);
     });
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
diff --git a/src/auto_schedule/measure_record.h b/src/auto_scheduler/measure_record.h
similarity index 91%
rename from src/auto_schedule/measure_record.h
rename to src/auto_scheduler/measure_record.h
index f97e30ae9f70..1cfeab07a400 100644
--- a/src/auto_schedule/measure_record.h
+++ b/src/auto_scheduler/measure_record.h
@@ -18,12 +18,12 @@
  */
 
 /*!
- * \file auto_schedule/measure_record.h
+ * \file auto_scheduler/measure_record.h
  * \brief Json serialization format for dumping and loading tuning records.
  */
 
-#ifndef TVM_AUTO_SCHEDULE_MEASURE_RECORD_H_
-#define TVM_AUTO_SCHEDULE_MEASURE_RECORD_H_
+#ifndef TVM_AUTO_SCHEDULER_MEASURE_RECORD_H_
+#define TVM_AUTO_SCHEDULER_MEASURE_RECORD_H_
 
 #include <fstream>
 #include <string>
@@ -32,7 +32,7 @@
 #include "measure.h"
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 /*! \brief Callback for logging the input and results of measurements to file */
 class RecordToFileNode : public MeasureCallbackNode {
@@ -43,7 +43,7 @@ class RecordToFileNode : public MeasureCallbackNode {
   void Callback(const SearchPolicy& policy, const Array<MeasureInput>& inputs,
                 const Array<MeasureResult>& results) final;
 
-  static constexpr const char* _type_key = "auto_schedule.RecordToFile";
+  static constexpr const char* _type_key = "auto_scheduler.RecordToFile";
   TVM_DECLARE_FINAL_OBJECT_INFO(RecordToFileNode, MeasureCallbackNode);
 };
 
@@ -88,7 +88,7 @@ class RecordReaderNode : public Object {
   std::pair<Array<MeasureInput>, Array<MeasureResult>> ReadLines(int max_size = -1,
                                                                  int skip_size = 0);
 
-  static constexpr const char* _type_key = "auto_schedule.RecordReader";
+  static constexpr const char* _type_key = "auto_scheduler.RecordReader";
   TVM_DECLARE_FINAL_OBJECT_INFO(RecordReaderNode, Object);
 
  private:
@@ -130,7 +130,7 @@ void WriteMeasureRecords(std::ostream* os, const Array<MeasureInput>& inputs,
 void ReadMeasureRecord(const std::string& str, MeasureInputNode* inp, MeasureResultNode* res,
                        std::string* log_version);
 
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
 
-#endif  // TVM_AUTO_SCHEDULE_MEASURE_RECORD_H_
+#endif  // TVM_AUTO_SCHEDULER_MEASURE_RECORD_H_
diff --git a/src/auto_schedule/search_policy/empty_policy.cc b/src/auto_scheduler/search_policy/empty_policy.cc
similarity index 94%
rename from src/auto_schedule/search_policy/empty_policy.cc
rename to src/auto_scheduler/search_policy/empty_policy.cc
index d91d563252b5..1886203593a9 100644
--- a/src/auto_schedule/search_policy/empty_policy.cc
+++ b/src/auto_scheduler/search_policy/empty_policy.cc
@@ -18,7 +18,7 @@
  */
 
 /*!
- * \file auto_schedule/search_policy/empty_policy.cc
+ * \file auto_scheduler/search_policy/empty_policy.cc
  * \brief This is an brief example of search policy.
  */
 
@@ -29,7 +29,7 @@
 #include "../measure.h"
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 TVM_REGISTER_NODE_TYPE(EmptyPolicyNode);
 
@@ -92,9 +92,9 @@ Array<State> EmptyPolicyNode::SearchOneRound() {
   return res;
 }
 
-TVM_REGISTER_GLOBAL("auto_schedule.EmptyPolicy").set_body_typed([]() {
+TVM_REGISTER_GLOBAL("auto_scheduler.EmptyPolicy").set_body_typed([]() {
   return EmptyPolicy(make_object<EmptyPolicyNode>());
 });
 
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
diff --git a/src/auto_schedule/search_policy/empty_policy.h b/src/auto_scheduler/search_policy/empty_policy.h
similarity index 85%
rename from src/auto_schedule/search_policy/empty_policy.h
rename to src/auto_scheduler/search_policy/empty_policy.h
index 610a02a3cd12..4ccc9c1042ea 100644
--- a/src/auto_schedule/search_policy/empty_policy.h
+++ b/src/auto_scheduler/search_policy/empty_policy.h
@@ -18,19 +18,19 @@
  */
 
 /*!
- * \file auto_schedule/search_policy/empty_policy.h
+ * \file auto_scheduler/search_policy/empty_policy.h
  * \brief A brief example of the search policy which always returns the initial naive schedule
  * (state).
  */
 
-#ifndef TVM_AUTO_SCHEDULE_SEARCH_POLICY_EMPTY_POLICY_H_
-#define TVM_AUTO_SCHEDULE_SEARCH_POLICY_EMPTY_POLICY_H_
+#ifndef TVM_AUTO_SCHEDULER_SEARCH_POLICY_EMPTY_POLICY_H_
+#define TVM_AUTO_SCHEDULER_SEARCH_POLICY_EMPTY_POLICY_H_
 
 #include "../loop_state.h"
 #include "search_policy.h"
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 /*!
  * \brief A brief example of the search policy which always returns the initial naive schedule
@@ -44,7 +44,7 @@ class EmptyPolicyNode : public SearchPolicyNode {
                int num_measures_per_round, int verbose, ProgramMeasurer measurer,
                Optional<Array<SearchCallback>> pre_search_callbacks) final;
 
-  static constexpr const char* _type_key = "auto_schedule.EmptyPolicy";
+  static constexpr const char* _type_key = "auto_scheduler.EmptyPolicy";
   TVM_DECLARE_FINAL_OBJECT_INFO(EmptyPolicyNode, SearchPolicyNode);
 
  private:
@@ -64,7 +64,7 @@ class EmptyPolicy : public SearchPolicy {
   TVM_DEFINE_MUTABLE_OBJECT_REF_METHODS(EmptyPolicy, SearchPolicy, EmptyPolicyNode);
 };
 
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
 
-#endif  // TVM_AUTO_SCHEDULE_SEARCH_POLICY_EMPTY_POLICY_H_
+#endif  // TVM_AUTO_SCHEDULER_SEARCH_POLICY_EMPTY_POLICY_H_
diff --git a/src/auto_schedule/search_policy/search_policy.cc b/src/auto_scheduler/search_policy/search_policy.cc
similarity index 84%
rename from src/auto_schedule/search_policy/search_policy.cc
rename to src/auto_scheduler/search_policy/search_policy.cc
index f8ac7ca39495..fba5155edaea 100644
--- a/src/auto_schedule/search_policy/search_policy.cc
+++ b/src/auto_scheduler/search_policy/search_policy.cc
@@ -18,7 +18,7 @@
  */
 
 /*!
- * \file auto_schedule/search_policy/search_policy.cc
+ * \file auto_scheduler/search_policy/search_policy.cc
  * \brief The base class of search policies.
  */
 
@@ -27,7 +27,7 @@
 #include <tvm/runtime/registry.h>
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 TVM_REGISTER_OBJECT_TYPE(SearchCallbackNode);
 TVM_REGISTER_OBJECT_TYPE(SearchPolicyNode);
@@ -40,16 +40,16 @@ void SearchPolicyNode::RunCallbacks(const Optional<Array<SearchCallback>>& callb
   }
 }
 
-TVM_REGISTER_GLOBAL("auto_schedule.SearchPolicyRunCallbacks")
+TVM_REGISTER_GLOBAL("auto_scheduler.SearchPolicyRunCallbacks")
     .set_body_typed([](SearchPolicy policy, Optional<Array<SearchCallback>> callbacks) {
       policy->RunCallbacks(callbacks);
     });
 
-TVM_REGISTER_GLOBAL("auto_schedule.SearchPolicySetTask")
+TVM_REGISTER_GLOBAL("auto_scheduler.SearchPolicySetTask")
     .set_body_typed([](SearchPolicy policy, SearchTask task) { policy->cur_task = task; });
 
-TVM_REGISTER_GLOBAL("auto_schedule.SearchPolicySetVerbose")
+TVM_REGISTER_GLOBAL("auto_scheduler.SearchPolicySetVerbose")
     .set_body_typed([](SearchPolicy policy, int verbose) { policy->verbose = verbose; });
 
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
diff --git a/src/auto_schedule/search_policy/search_policy.h b/src/auto_scheduler/search_policy/search_policy.h
similarity index 93%
rename from src/auto_schedule/search_policy/search_policy.h
rename to src/auto_scheduler/search_policy/search_policy.h
index 47cccec93661..70f94ad65b94 100644
--- a/src/auto_schedule/search_policy/search_policy.h
+++ b/src/auto_scheduler/search_policy/search_policy.h
@@ -18,7 +18,7 @@
  */
 
 /*!
- * \file auto_schedule/search_policy/search_policy.h
+ * \file auto_scheduler/search_policy/search_policy.h
  * \brief The base class of search policies, including the abstract definition of search policy and
  * other supporting data structures.
  *
@@ -48,8 +48,8 @@
  * during the search process.
  */
 
-#ifndef TVM_AUTO_SCHEDULE_SEARCH_POLICY_SEARCH_POLICY_H_
-#define TVM_AUTO_SCHEDULE_SEARCH_POLICY_SEARCH_POLICY_H_
+#ifndef TVM_AUTO_SCHEDULER_SEARCH_POLICY_SEARCH_POLICY_H_
+#define TVM_AUTO_SCHEDULER_SEARCH_POLICY_SEARCH_POLICY_H_
 
 #include <tvm/node/node.h>
 
@@ -59,7 +59,7 @@
 #include "../search_task.h"
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 class ProgramMeasurer;
 class SearchPolicyNode;
@@ -77,7 +77,7 @@ class SearchCallbackNode : public Object {
    */
   virtual void Callback(SearchPolicyNode* policy) = 0;
 
-  static constexpr const char* _type_key = "auto_schedule.SearchCallback";
+  static constexpr const char* _type_key = "auto_scheduler.SearchCallback";
   TVM_DECLARE_BASE_OBJECT_INFO(SearchCallbackNode, Object);
 };
 
@@ -131,7 +131,7 @@ class SearchPolicyNode : public Object {
    */
   void RunCallbacks(const Optional<Array<SearchCallback>>& callbacks);
 
-  static constexpr const char* _type_key = "auto_schedule.SearchPolicy";
+  static constexpr const char* _type_key = "auto_scheduler.SearchPolicy";
   TVM_DECLARE_BASE_OBJECT_INFO(SearchPolicyNode, Object);
 
  protected:
@@ -161,7 +161,7 @@ class SearchPolicy : public ObjectRef {
   TVM_DEFINE_MUTABLE_OBJECT_REF_METHODS(SearchPolicy, ObjectRef, SearchPolicyNode);
 };
 
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
 
-#endif  // TVM_AUTO_SCHEDULE_SEARCH_POLICY_SEARCH_POLICY_H_
+#endif  // TVM_AUTO_SCHEDULER_SEARCH_POLICY_SEARCH_POLICY_H_
diff --git a/src/auto_schedule/search_task.cc b/src/auto_scheduler/search_task.cc
similarity index 93%
rename from src/auto_schedule/search_task.cc
rename to src/auto_scheduler/search_task.cc
index 1d7a08cc73db..912a31046540 100644
--- a/src/auto_schedule/search_task.cc
+++ b/src/auto_scheduler/search_task.cc
@@ -18,7 +18,7 @@
  */
 
 /*!
- * \file auto_schedule/search_task.cc
+ * \file auto_scheduler/search_task.cc
  * \brief Meta information and hardware parameters for a search task.
  */
 
@@ -30,7 +30,7 @@
 #include <utility>
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 TVM_REGISTER_NODE_TYPE(HardwareParamsNode);
 TVM_REGISTER_NODE_TYPE(SearchTaskNode);
@@ -69,16 +69,16 @@ SearchTask::SearchTask(ComputeDAG compute_dag, String workload_key, Target targe
   data_ = std::move(node);
 }
 
-TVM_REGISTER_GLOBAL("auto_schedule.HardwareParams")
+TVM_REGISTER_GLOBAL("auto_scheduler.HardwareParams")
     .set_body_typed([](int num_cores, int vector_unit_bytes, int cache_line_bytes) {
       return HardwareParams(num_cores, vector_unit_bytes, cache_line_bytes);
     });
 
-TVM_REGISTER_GLOBAL("auto_schedule.SearchTask")
+TVM_REGISTER_GLOBAL("auto_scheduler.SearchTask")
     .set_body_typed([](ComputeDAG compute_dag, String workload_key, Target target,
                        Target target_host, Optional<HardwareParams> hardware_params) {
       return SearchTask(compute_dag, workload_key, target, target_host, hardware_params);
     });
 
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
diff --git a/src/auto_schedule/search_task.h b/src/auto_scheduler/search_task.h
similarity index 92%
rename from src/auto_schedule/search_task.h
rename to src/auto_scheduler/search_task.h
index c7d2ddc533ed..ca313500cc8f 100644
--- a/src/auto_schedule/search_task.h
+++ b/src/auto_scheduler/search_task.h
@@ -18,19 +18,19 @@
  */
 
 /*!
- * \file auto_schedule/search_task.h
+ * \file auto_scheduler/search_task.h
  * \brief Meta information and hardware parameters for a search task.
  */
 
-#ifndef TVM_AUTO_SCHEDULE_SEARCH_TASK_H_
-#define TVM_AUTO_SCHEDULE_SEARCH_TASK_H_
+#ifndef TVM_AUTO_SCHEDULER_SEARCH_TASK_H_
+#define TVM_AUTO_SCHEDULER_SEARCH_TASK_H_
 
 #include <tvm/target/target.h>
 
 #include "compute_dag.h"
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 class HardwareParams;
 
@@ -76,7 +76,7 @@ class HardwareParamsNode : public Object {
    */
   static HardwareParams GetDefaultHardwareParams(const Target& target, const Target& target_host);
 
-  static constexpr const char* _type_key = "auto_schedule.HardwareParams";
+  static constexpr const char* _type_key = "auto_scheduler.HardwareParams";
   TVM_DECLARE_FINAL_OBJECT_INFO(HardwareParamsNode, Object);
 };
 
@@ -122,7 +122,7 @@ class SearchTaskNode : public Object {
     v->Visit("hardware_params", &hardware_params);
   }
 
-  static constexpr const char* _type_key = "auto_schedule.SearchTask";
+  static constexpr const char* _type_key = "auto_scheduler.SearchTask";
   TVM_DECLARE_FINAL_OBJECT_INFO(SearchTaskNode, Object);
 };
 
@@ -146,7 +146,7 @@ class SearchTask : public ObjectRef {
   TVM_DEFINE_OBJECT_REF_METHODS(SearchTask, ObjectRef, SearchTaskNode);
 };
 
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
 
-#endif  // TVM_AUTO_SCHEDULE_SEARCH_TASK_H_
+#endif  // TVM_AUTO_SCHEDULER_SEARCH_TASK_H_
diff --git a/src/auto_schedule/transform_step.cc b/src/auto_scheduler/transform_step.cc
similarity index 98%
rename from src/auto_schedule/transform_step.cc
rename to src/auto_scheduler/transform_step.cc
index bffb2dcfab31..90b4db838fef 100644
--- a/src/auto_schedule/transform_step.cc
+++ b/src/auto_scheduler/transform_step.cc
@@ -18,7 +18,7 @@
  */
 
 /*!
- * \file auto_schedule/transform_step.cc
+ * \file auto_scheduler/transform_step.cc
  * \brief Transformation steps. For each schedule primitive, there is a corresponding transform
  * step.
  */
@@ -34,7 +34,7 @@
 #include "utils.h"
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 /********** Reorder **********/
 ReorderStep::ReorderStep(int stage_id, const Array<Integer>& after_ids) {
@@ -235,5 +235,5 @@ String FuseStepNode::PrintAsPythonAPI(Array<te::Stage>* stages,
   return ss.str();
 }
 
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
diff --git a/src/auto_schedule/transform_step.h b/src/auto_scheduler/transform_step.h
similarity index 92%
rename from src/auto_schedule/transform_step.h
rename to src/auto_scheduler/transform_step.h
index 5e54c9de583f..d840cc009e2d 100644
--- a/src/auto_schedule/transform_step.h
+++ b/src/auto_scheduler/transform_step.h
@@ -18,7 +18,7 @@
  */
 
 /*!
- * \file auto_schedule/transform_step.h
+ * \file auto_scheduler/transform_step.h
  * \brief Transformation steps. For each schedule primitive, there is a corresponding transform
  * step. The implementation of each step consists of 2 parts:
  * - transform_step.cc: How each step interacts with TE and TE's schedule primitives
@@ -34,13 +34,13 @@
  *    - In these two functions you need to incrementally update all data structures in State with
  *      CopyOnWrite style
  * 4. Add you step to `ComputeDAG::ApplySteps` and make sure it works.
- * 5. Add log record serialization support in `struct Handler<Array<::tvm::auto_schedule::Step>>`
+ * 5. Add log record serialization support in `struct Handler<Array<::tvm::auto_scheduler::Step>>`
  *    in `record.cc`.
  * 6. Add its corresponding Python API to `loop_state.py` and necessary unit test.
  */
 
-#ifndef TVM_AUTO_SCHEDULE_TRANSFORM_STEP_H_
-#define TVM_AUTO_SCHEDULE_TRANSFORM_STEP_H_
+#ifndef TVM_AUTO_SCHEDULER_TRANSFORM_STEP_H_
+#define TVM_AUTO_SCHEDULER_TRANSFORM_STEP_H_
 
 #include <dmlc/common.h>
 #include <tvm/node/node.h>
@@ -49,7 +49,7 @@
 #include "utils.h"
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 typedef Map<tvm::te::Stage, Array<tir::IterVar>, ObjectHash, ObjectEqual> StageToAxesMap;
 
@@ -62,7 +62,7 @@ class StepNode : public Object {
   /*! \brief The index of the stage. */
   int stage_id;
 
-  static constexpr const char* _type_key = "auto_schedule.Step";
+  static constexpr const char* _type_key = "auto_scheduler.Step";
   TVM_DECLARE_BASE_OBJECT_INFO(StepNode, Object);
 };
 
@@ -99,7 +99,7 @@ class ReorderStepNode : public StepNode {
    */
   String PrintAsPythonAPI(Array<te::Stage>* stages, StageToAxesMap* stage_to_axes) const;
 
-  static constexpr const char* _type_key = "auto_schedule.ReorderStep";
+  static constexpr const char* _type_key = "auto_scheduler.ReorderStep";
   TVM_DECLARE_FINAL_OBJECT_INFO(ReorderStepNode, Object);
 };
 
@@ -154,7 +154,7 @@ class SplitStepNode : public StepNode {
    */
   String PrintAsPythonAPI(Array<te::Stage>* stages, StageToAxesMap* stage_to_axes) const;
 
-  static constexpr const char* _type_key = "auto_schedule.SplitStep";
+  static constexpr const char* _type_key = "auto_scheduler.SplitStep";
   TVM_DECLARE_FINAL_OBJECT_INFO(SplitStepNode, Object);
 };
 
@@ -200,7 +200,7 @@ class FuseStepNode : public StepNode {
    */
   String PrintAsPythonAPI(Array<te::Stage>* stages, StageToAxesMap* stage_to_axes) const;
 
-  static constexpr const char* _type_key = "auto_schedule.FuseStep";
+  static constexpr const char* _type_key = "auto_scheduler.FuseStep";
   TVM_DECLARE_FINAL_OBJECT_INFO(FuseStepNode, Object);
 };
 
@@ -220,7 +220,7 @@ class FuseStep : public Step {
   TVM_DEFINE_OBJECT_REF_METHODS(FuseStep, Step, FuseStepNode);
 };
 
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
 
-#endif  // TVM_AUTO_SCHEDULE_TRANSFORM_STEP_H_
+#endif  // TVM_AUTO_SCHEDULER_TRANSFORM_STEP_H_
diff --git a/src/auto_schedule/utils.cc b/src/auto_scheduler/utils.cc
similarity index 91%
rename from src/auto_schedule/utils.cc
rename to src/auto_scheduler/utils.cc
index ecb6145268d6..68f503836cfb 100644
--- a/src/auto_schedule/utils.cc
+++ b/src/auto_scheduler/utils.cc
@@ -18,19 +18,19 @@
  */
 
 /*!
- * \file auto_schedule/utils.cc
+ * \file auto_scheduler/utils.cc
  * \brief Common utilities.
  */
 
 #include "utils.h"
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 NullStream& NullStream::Global() {
   static NullStream stream;
   return stream;
 }
 
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
diff --git a/src/auto_schedule/utils.h b/src/auto_scheduler/utils.h
similarity index 96%
rename from src/auto_schedule/utils.h
rename to src/auto_scheduler/utils.h
index e91bc106fb51..5637780e3991 100644
--- a/src/auto_schedule/utils.h
+++ b/src/auto_scheduler/utils.h
@@ -18,12 +18,12 @@
  */
 
 /*!
- * \file auto_schedule/utils.h
+ * \file auto_scheduler/utils.h
  * \brief Common utilities.
  */
 
-#ifndef TVM_AUTO_SCHEDULE_UTILS_H_
-#define TVM_AUTO_SCHEDULE_UTILS_H_
+#ifndef TVM_AUTO_SCHEDULER_UTILS_H_
+#define TVM_AUTO_SCHEDULER_UTILS_H_
 
 #include <dmlc/common.h>
 #include <tvm/tir/expr.h>
@@ -61,7 +61,7 @@ struct hash<std::tuple<T1, T2, T3>> {
 }  // namespace std
 
 namespace tvm {
-namespace auto_schedule {
+namespace auto_scheduler {
 
 /********** Utilities for Array, std::string **********/
 /*! \brief Get the first appearance index of elements in an Array */
@@ -183,7 +183,7 @@ inline void PrintTitle(const std::string& title, int verbose) {
                    << Chars('-', 60) << std::endl;
 }
 
-}  // namespace auto_schedule
+}  // namespace auto_scheduler
 }  // namespace tvm
 
-#endif  // TVM_AUTO_SCHEDULE_UTILS_H_
+#endif  // TVM_AUTO_SCHEDULER_UTILS_H_
diff --git a/tests/python/unittest/test_auto_schedule_common.py b/tests/python/unittest/test_auto_scheduler_common.py
similarity index 88%
rename from tests/python/unittest/test_auto_schedule_common.py
rename to tests/python/unittest/test_auto_scheduler_common.py
index 691c7e767b6f..078e1ae8e854 100644
--- a/tests/python/unittest/test_auto_schedule_common.py
+++ b/tests/python/unittest/test_auto_scheduler_common.py
@@ -15,16 +15,16 @@
 # specific language governing permissions and limitations
 # under the License.
 
-"""Common functions for auto_schedule test cases"""
+"""Common functions for auto_scheduler test cases"""
 
 import threading
 
-from tvm import te, auto_schedule
+from tvm import te, auto_scheduler
 import topi
 
 
-@auto_schedule.register_workload
-def matmul_auto_schedule_test(N, M, K):
+@auto_scheduler.register_workload
+def matmul_auto_scheduler_test(N, M, K):
     A = te.placeholder((N, K), name='A')
     B = te.placeholder((K, M), name='B')
     k = te.reduce_axis((0, K), name='k')
@@ -32,8 +32,8 @@ def matmul_auto_schedule_test(N, M, K):
     return [A, B, C]
 
 
-@auto_schedule.register_workload("matmul_auto_schedule_test_rename_1")
-def matmul_auto_schedule_test_rename_0(N, M, K):
+@auto_scheduler.register_workload("matmul_auto_scheduler_test_rename_1")
+def matmul_auto_scheduler_test_rename_0(N, M, K):
     A = te.placeholder((N, K), name='A')
     B = te.placeholder((K, M), name='B')
     k = te.reduce_axis((0, K), name='k')
@@ -67,8 +67,8 @@ def conv2d_nchw_bn_relu(N, H, W, CI, CO, kernel_size, strides, padding, dilation
 
 
 def get_tiled_matmul():
-    A, B, C = matmul_auto_schedule_test(512, 512, 512)
-    dag = auto_schedule.ComputeDAG([A, B, C])
+    A, B, C = matmul_auto_scheduler_test(512, 512, 512)
+    dag = auto_scheduler.ComputeDAG([A, B, C])
 
     s0 = dag.get_init_state()
     its0 = s0.split(C, s0[C].iters[0], [4, 8, 8])
diff --git a/tests/python/unittest/test_auto_schedule_compute_dag.py b/tests/python/unittest/test_auto_scheduler_compute_dag.py
similarity index 93%
rename from tests/python/unittest/test_auto_schedule_compute_dag.py
rename to tests/python/unittest/test_auto_scheduler_compute_dag.py
index 8a4f836765eb..49344634a8f0 100644
--- a/tests/python/unittest/test_auto_schedule_compute_dag.py
+++ b/tests/python/unittest/test_auto_scheduler_compute_dag.py
@@ -18,9 +18,9 @@
 """Test ComputeDAG (replay, infer bound)"""
 
 import tvm
-from tvm import auto_schedule, te
+from tvm import auto_scheduler, te
 
-from test_auto_schedule_common import get_tiled_matmul
+from test_auto_scheduler_common import get_tiled_matmul
 
 
 def test_apply_steps():
diff --git a/tests/python/unittest/test_auto_schedule_loop_state.py b/tests/python/unittest/test_auto_scheduler_loop_state.py
similarity index 89%
rename from tests/python/unittest/test_auto_schedule_loop_state.py
rename to tests/python/unittest/test_auto_scheduler_loop_state.py
index ed54da513d16..0801d9200275 100644
--- a/tests/python/unittest/test_auto_schedule_loop_state.py
+++ b/tests/python/unittest/test_auto_scheduler_loop_state.py
@@ -20,15 +20,15 @@
 import numpy as np
 
 import tvm
-from tvm import auto_schedule, te
+from tvm import auto_scheduler, te
 import topi
 
-from test_auto_schedule_common import matmul_auto_schedule_test, conv2d_nchw_bn_relu
+from test_auto_scheduler_common import matmul_auto_scheduler_test, conv2d_nchw_bn_relu
 
 
 def test_split_fuse_reorder():
-    A, B, C = matmul_auto_schedule_test(512, 512, 512)
-    dag = auto_schedule.ComputeDAG([A, B, C])
+    A, B, C = matmul_auto_scheduler_test(512, 512, 512)
+    dag = auto_scheduler.ComputeDAG([A, B, C])
     s0 = dag.get_init_state()
     i, j, k = s0[C].iters
 
diff --git a/tests/python/unittest/test_auto_schedule_measure.py b/tests/python/unittest/test_auto_scheduler_measure.py
similarity index 74%
rename from tests/python/unittest/test_auto_schedule_measure.py
rename to tests/python/unittest/test_auto_scheduler_measure.py
index 52d016de0756..1bcd0540ecb4 100644
--- a/tests/python/unittest/test_auto_schedule_measure.py
+++ b/tests/python/unittest/test_auto_scheduler_measure.py
@@ -18,10 +18,10 @@
 """ Test measurement and log serialization. """
 
 import tvm
-from tvm import auto_schedule
+from tvm import auto_scheduler
 import tempfile
 
-from test_auto_schedule_common import get_tiled_matmul
+from test_auto_scheduler_common import get_tiled_matmul
 
 
 def test_record():
@@ -30,15 +30,15 @@ def test_record():
     if not tvm.runtime.enabled("llvm"):
         return
     target = tvm.target.create("llvm")
-    task = auto_schedule.SearchTask(dag, "test", target)
+    task = auto_scheduler.SearchTask(dag, "test", target)
 
-    inp = auto_schedule.measure.MeasureInput(task, s)
-    res = auto_schedule.measure.MeasureResult([0.1], 0, "", 0.2, 1)
+    inp = auto_scheduler.measure.MeasureInput(task, s)
+    res = auto_scheduler.measure.MeasureResult([0.1], 0, "", 0.2, 1)
 
     with tempfile.NamedTemporaryFile() as fp:
-        auto_schedule.save_records(fp.name, [inp], [res])
+        auto_scheduler.save_records(fp.name, [inp], [res])
 
-        log_reader = auto_schedule.RecordReader(fp.name)
+        log_reader = auto_scheduler.RecordReader(fp.name)
         inputs, results = log_reader.read_lines()
         assert len(inputs) == 1
 
@@ -55,11 +55,11 @@ def test_measure_local_builder_runner():
     if not tvm.runtime.enabled("llvm"):
         return
     tgt = tvm.target.create("llvm")
-    task = auto_schedule.SearchTask(dag, "test", tgt)
+    task = auto_scheduler.SearchTask(dag, "test", tgt)
 
-    minp = auto_schedule.MeasureInput(task, s0)
-    local_builder = auto_schedule.LocalBuilder()
-    local_runner = auto_schedule.LocalRunner()
+    minp = auto_scheduler.MeasureInput(task, s0)
+    local_builder = auto_scheduler.LocalBuilder()
+    local_runner = auto_scheduler.LocalRunner()
 
     bress = local_builder.build([minp])
     assert bress[0].error_no == 0
diff --git a/tests/python/unittest/test_auto_schedule_search_policy.py b/tests/python/unittest/test_auto_scheduler_search_policy.py
similarity index 78%
rename from tests/python/unittest/test_auto_schedule_search_policy.py
rename to tests/python/unittest/test_auto_scheduler_search_policy.py
index 9e08218dcbce..8dfc07865ea8 100644
--- a/tests/python/unittest/test_auto_schedule_search_policy.py
+++ b/tests/python/unittest/test_auto_scheduler_search_policy.py
@@ -22,31 +22,32 @@
 import tempfile
 
 import tvm
-from tvm import auto_schedule
+from tvm import auto_scheduler
 
-from test_auto_schedule_common import matmul_auto_schedule_test, PropagatingThread
+from test_auto_scheduler_common import matmul_auto_scheduler_test, PropagatingThread
 
-def search_common(workload=matmul_auto_schedule_test, target="llvm", search_policy = auto_schedule.EmptyPolicy(),
+def search_common(workload=matmul_auto_scheduler_test, target="llvm",
+                  search_policy=auto_scheduler.EmptyPolicy(),
                   seed=random.randint(1, 1 << 30), runner='local', cost_model=None,
                   num_measure_trials=2, params=None, pre_search_callbacks=None):
     print("Test %s schedule search with the default search policy" % (target))
 
     random.seed(seed)
     N = 128
-    workload_key = auto_schedule.make_workload_key(workload, (N, N, N))
-    dag = auto_schedule.ComputeDAG(workload_key)
+    workload_key = auto_scheduler.make_workload_key(workload, (N, N, N))
+    dag = auto_scheduler.ComputeDAG(workload_key)
     target = tvm.target.create(target)
-    task = auto_schedule.SearchTask(dag, workload_key, target)
+    task = auto_scheduler.SearchTask(dag, workload_key, target)
 
     with tempfile.NamedTemporaryFile() as fp:
         log_file = fp.name
 
-        tuning_options = auto_schedule.TuningOptions(num_measure_trials=num_measure_trials, runner=runner,
+        tuning_options = auto_scheduler.TuningOptions(num_measure_trials=num_measure_trials, runner=runner,
                                              verbose=0,
-                                             measure_callbacks=[auto_schedule.RecordToFile(log_file)],
+                                             measure_callbacks=[auto_scheduler.RecordToFile(log_file)],
                                              pre_search_callbacks=pre_search_callbacks)
-        sch, args = auto_schedule.auto_schedule(task, search_policy, tuning_options)
-        inp, res = auto_schedule.load_best(log_file, workload_key, target)
+        sch, args = auto_scheduler.auto_schedule(task, search_policy, tuning_options)
+        inp, res = auto_scheduler.load_best(log_file, workload_key, target)
 
         print("==== Python Code ====")
         print(dag.print_python_code_from_state(inp.state))
@@ -79,11 +80,11 @@ def test_workload_registry_search_basic():
     t.start()
     t.join()
     t = PropagatingThread(target=search_common,
-                          kwargs={'seed': 944563397, 'workload': "matmul_auto_schedule_test"})
+                          kwargs={'seed': 944563397, 'workload': "matmul_auto_scheduler_test"})
     t.start()
     t.join()
     t = PropagatingThread(target=search_common,
-                          kwargs={'seed': 944563397, 'workload': "matmul_auto_schedule_test_rename_1"})
+                          kwargs={'seed': 944563397, 'workload': "matmul_auto_scheduler_test_rename_1"})
     t.start()
     t.join()