Applied AI for Research

Self-Hosted Bioinformatics AI / Internal RAG System

A local research assistant built around retrieval-augmented generation for curated scientific content, with a focus on internal use rather than generic chatbot behavior.

MSDS Capstone Project Project Lead January 2025 - May 2025

What it does

The goal was to build a lab-oriented assistant that could retrieve from curated scientific material locally, rather than rely on a generic external chat workflow. That meant prioritizing ingestion, chunking, indexing, retrieval quality, and practical usability for research contexts.

It was designed as a tool for controlled research environments where curated corpora and offline or local workflows can matter more than broad general-purpose chat behavior.

Core stack

Document ingestion and text chunking
Sentence-transformer embeddings
FAISS vector indexing
Retrieval workflows
Streamlit interface

What I built

Led the project concept and a substantial share of implementation
Built ingestion and chunking workflows for curated scientific material
Implemented vector indexing and retrieval logic for lab-facing querying
Helped shape the project around practical internal research use

Availability

This was an MSDS capstone project and is summarized here as representative work. The page focuses on the architecture and intent rather than exposing a raw code dump.