At HSBC Electronic Data Processing India Pvt Ltd, I built an enterprise-grade RAG system for banking compliance documents, improving retrieval relevance by 30%. I also implemented citation-backed answers to make document search more context-aware and explainable.
At PricewaterhouseCoopers India, I built fraud detection models for a banking client that achieved an F1 score of 0.95 and reduced false positives in production workflows. I also designed oil-industry predictive and optimization models that improved cash flow by 35% and delivered 8× cost savings.
On Mutual Fund Prospectus Analyser, I built a RAG-based document intelligence platform and chatbot for extracting and analyzing information from unstructured prospectuses. The project used GPT-4o, LangChain, and ChromaDB for semantic search and citation-backed responses.
On RegX Horizon Scanner, I developed a multi-agent regulatory intelligence platform using LangGraph, LangChain, Claude, and ChromaDB. I also built an A2A-enabled RAG pipeline supporting collaboration across six AI agents.
