AI Systems & Backend Engineer focused on controlled AI execution, production infrastructure, backend systems, automation and verification.
I build and operate systems where AI can perform substantial implementation work while scope, state, verification, recovery and operational outcomes remain controlled.
This GitHub profile is the public evidence layer behind that work: small inspectable artifacts, defensive engineering patterns, public client code and historical prototypes that can be reviewed directly.
This profile is a current public workbench, not a long-standing open-source track record. The repositories are recent and should be evaluated as code samples.
Public thin client for RailTaskLite, a controlled AI execution runtime built around bounded work, explicit state, verification, recovery and server-authoritative licensing. Supports Claude Code / Codex, with public security regression tests and CI.
Tested, read-only system inspection and command-center tooling.
Focused backend, security, and reliability samples.
Includes HMAC metadata signing, atomic file writes, safe upload handling, cursor pagination, runtime drift records, state transition validation, and audio timing maps.
Installable image optimization and upscaling tool using Real-ESRGAN, with CI.
Includes a local web UI and practical image-processing workflow.
Historical/experimental orchestration prototype, an early predecessor to RailTaskLite. Not a production system.
Small side experiment: local terminal/operator tooling for project-file injection, code generation, and session persistence.
- controlled AI execution
- backend systems
- Linux infrastructure
- automation and orchestration
- verification and recovery
- production debugging
- CLI workflows
- testing and reliability
- security-conscious system design
- Docker
- PostgreSQL
- Redis
- FastAPI
- Python-based systems
For the strongest current product evidence, start with the RailTaskLite client. For focused backend, reliability and defensive-engineering examples, review production-engineering-samples.
I do not present this account as evidence of long-term open-source community activity. It is a recent public portfolio of small, inspectable artifacts.
Website: alexbachlej.com