On EvalForge, I built a GitHub Actions pipeline that runs golden Q&A cases against a RAG system on every commit, surfacing hallucinations and retrieval failures before merge. I used RAGAS to track evaluation scores and response latency, with results stored in ChromaDB.
I also implemented threshold-based CI quality gates and developed a Streamlit dashboard to compare prompt versions, model configurations, and retrieval settings over time.
As a contributor to Dream11 OTA (topoteretes/cognee), I resolved two production bugs and implemented backend access-control logic and a frontend tenant-header helper. For ShopSphere, I created a reusable React component library and achieved full component-level test coverage with Jest and React Testing Library.
