Soham M.

SHL Assessment Recommender

Built and deployed a Retrieval-Augmented Generation (RAG) application using hybrid retrieval (FAISS semantic search, 70% weight + BM25 keyword search, 30% weight) grounded into an LLM (Groq, Llama 3.3 70B).

AI Powered Image Retrieval

Deployed a scalable image retrieval app (FAISS, Docker) with sub-0.2s latency, improving demo efficiency for stakeholders by ~70%.