At TATA Consultancy Services, I engineered an on-demand RAG system for medical product experience risk analysis, processing thousands of risk documents. Using AzureOpenAI embeddings and semantic search through FastAPI, the system delivered context-aware analysis with 1–2 second standard endpoint latency.
I architected a curricula recommendation engine using a 60GB OpenSearch vector index, with nested vectors for training content from about 33,000 curricula and 160,000 training items. It delivered specific, context-aware learning paths.
I also built end-to-end tracing with Langfuse for a healthcare code retrieval system, monitoring model latency and token usage. Earlier, I developed PowerShell utilities for Active Directory group management and Azure log retrieval, and engineered an automated build solution for an enterprise POS system.
