Building AI-ready platforms, event-driven infrastructure, and data-enabled products that scale to 500+ applications and $48M+ in revenue.
A production-grade Retrieval-Augmented Generation system that ingests PDFs, Confluence, and Slack data — chunks, embeds, and stores in a vector DB — then provides conversational search with real-time cited answers and streaming responses.
Built a RAG-based enterprise knowledge search engine using LangChain, FastAPI, and ChromaDB, enabling semantic search over 10K+ documents with hybrid retrieval (vector + BM25), cross-encoder re-ranking, and real-time cited responses — reducing information retrieval time by 80%.
LSTM Neural Network with sentiment analysis via Python NLTK Vader lexicon, predicting 60-day stock price momentum using time series forecasting.
Restaurant recommendation system via sentiment analysis on 50,000 customer reviews — TF-IDF feature extraction translates raw text into actionable numeric ratings.