I have been building and maintaining production software for many years, primarily in domains where reliability, traceability, and long-term maintainability matter (industrial automation, data pipelines, decision-support systems).

My current focus is RAE (Reflective Agentic-Memory Engine), an open-source project that treats memory as a governed subsystem for AI agents rather than a simple vector store or chat history. RAE introduces layered memory, explicit retention and consolidation rules, and auditability/provenance as first-class concepts.

The project grew out of practical frustration with prompt drift, opaque reasoning, and lack of reproducibility in agentic systems. My goal is to build reusable, local-first infrastructure that others can audit, extend, and rely on — not a closed product or demo.