I've built RAG-powered AI products that turn unstructured data and user behavior into personalized guidance, controls, and feedback.
At Swiss Re, I built an end-to-end career guidance chatbot on Palantir Foundry that transformed resumes into personalized skill-gap analysis. I designed section-aware chunking, semantic search, query rewriting, and token-cost estimation in a custom Python backend.
My projects include Breakthrough, an on-device AI speaking coach with real-time transcription, prosody analysis, voice-activity detection, and history-aware feedback. I also built Syso, a live voice-driven system caretaker that uses multimodal interaction and RAG to personalize diagnostics, alerts, and settings management.
I created a MongoDB index-optimization tool that automated detection, optimization, and performance profiling, reducing query time and storage costs by 30% while identifying redundant indexes with 95% accuracy. I also led Team Sleepcode to first place at NIT Trichy's Transfinitte'24 hackathon.
