At Tata Consultancy Services, I architected a LangGraph conversational Q&A agent on Azure OpenAI for 400+ CPG analysts. It processed 5K+ daily queries and improved SQL accuracy from 71% to 92% with safety guardrails and multi-turn follow-up resolution.
I also built a multimodal RAG platform for video and image retrieval, improving precision@5 by 60% while maintaining sub-500ms p99 latency. Across my work, I focus on making production GenAI systems reliable through retrieval, evaluation, and deployment practices.
