I work with data long enough for it to fall into rhythm.
I am a Data Science undergraduate specializing in NLP, Retrieval Augmented Generation (RAG) systems, and data pipelines. I build retrieval centric architectures with a focus on evaluation methodology, accuracy, and deployment constraints.
Built a self hosted cloud storage system with an integrated RAG pipeline for querying technical PDFs. Implemented a retrieval strategy using hybrid search, re ranking, and context sufficiency checks to prevent hallucinations. Features asynchronous document ingestion and citation tracking, optimizing the system for faithfulness and measurable retrieval accuracy.
Automated event assistant that syncs Google Sheets data every 5 minutes for real-time knowledge updates. Handles chunking and Q&A with 663ms average latency, stress-tested with 30-50 concurrent users.
Full-stack web app for real-time news sentiment analysis. Scrapes articles, calculates VADER polarity scores, and classifies topics using a custom NLP pipeline.
Built FY24–FY25 insurance analytics pipelines (34 insurers, 9 segments) with validated YTD growth and risk metrics, and a hybrid retrieval RAG system over 52 documents with a Streamlit interface.
Contributor to Keras — merged pull request in core ML ecosystem
