AI Engineer from Bengaluru 🇮🇳 who turns raw sensor data into decisions. My work lives where industrial IT‑OT pipelines, generative AI and production MLOps meet.

• 🏭 Industrial AI — end‑to‑end IT‑OT pipelines ingesting high‑frequency engine/plant sensor data over raw TCP/IP, streamed to live dashboards over WebSockets
• 🧠 GenAI in production — RAG over technical test reports (LangChain · ChromaDB · rerankers) and agentic natural‑language‑to‑SQL workflows with LangGraph and MCP
• 🔭 Building — inferno: distributed ML inference (FastAPI · Redis Streams · dynamic batching · KEDA autoscaling)
• 📈 Deep learning — LSTM time‑series for anomaly detection and predictive maintenance; LiDAR + camera fusion and 3D object detection on KITTI
• ⚙️ MLOps — Docker + Kubernetes (KEDA autoscaling), CI/CD with GitHub Actions, Prometheus + OpenTelemetry, eval regression gates in CI
• 🌱 Learning — distributed systems depth, model‑serving internals, and the maths under all of it
• 👯 Open to collaborate on — production AI systems, industrial/IIoT and GenAI platforms
• 💬 Ask me about — RAG pipelines, sensor calibration, and shipping models that don't fall over in prod
• ⚡ Fun fact — Change is the only constant.

Production‑grade distributed ML inference — FastAPI gateway, Redis‑backed queue, dynamic batching and horizontally scaled workers.

A depth reference for distributed systems — 123 pages across 21 sections, 325 diagrams, runnable implementations, and CI that validates every link, scene and generated file.

Interactive RAG internals playground — chunking, BM25, ColBERT MaxSim, cross‑encoder reranking and RRF fusion, all computed live in the browser on your own text. No backend.

The 45‑minute interview round — a repeatable framework, the building blocks, and 8 worked designs with the trade‑offs stated out loud.

FPN‑ResNet + YOLO fusion for 3D object detection across LiDAR point clouds and camera frames on KITTI.

Building, testing and operating LLM systems — RAG, evaluation, serving, agents — each topic at three depths and from seven seats.

• Scale inferno — GPU workers, autoscaling
• Multi‑sensor 3D detection benchmarks
• Statistics for ML, deeper math

• ONNX / TensorRT edge deployment
• Multimodal RAG (vision + text)
• Kaggle competition medal 🥉

• Publish a perception paper / blog series
• Contribute to an open‑source CV library
• Mentor & collaborate on AI apps