GenAI & RAG Engineer · Python · LangChain · Vector Databases · Azure OpenAI
I build retrieval-augmented generation (RAG) systems and care about making them measurable: retrieval quality, latency and cost, not just demos.
| Project | What it does | Stack |
|---|---|---|
| RAG-Document-QA-Pipeline | PDF ingestion → Chroma vector store → grounded answers with sources | LangChain, ChromaDB, HuggingFace, OpenAI |
| MOPD-RAG | A router sends each question to specialist "teacher" models; a distiller merges their answers. Custom TF-IDF retriever | Python, Anthropic API |
| sentinel-rag | Agentic RAG blueprint for extracting data from messy procurement PDFs | LangChain, PDF parsing |
| InsightVanillaRAG | Baseline enterprise RAG on Azure OpenAI, the reference point for retrieval improvements | Azure OpenAI, FAISS |
| ResolveIQ | Customer-support copilot focused on ticket deflection and the cost-latency-quality trade-off | RAG |
Python LangChain ChromaDB FAISS Pinecone Azure OpenAI Anthropic API HuggingFace Power BI SQL Git
- Adding evaluation (retrieval hit rate, faithfulness, latency) to my RAG projects
- Turning prototypes into deployable APIs with tests and CI