At Xelron, I design realistic debugging and software development scenarios to expose edge cases in AI coding agents. I build reproducible benchmarks and Docker-based execution environments, with automated validation using Python, Bash, Docker, and PyTest.
I also built VedaAI, a full-stack assessment platform using Gemini and Claude APIs, with deterministic answer mapping and live progress streaming. My projects include a FastAPI support ticket platform and an on-device healthcare RAG assistant built with LangChain, Ollama, and ChromaDB.

