I build search, retrieval, and NLP systems — some at work, some just because I wanted to know how they worked. Currently a Master's student in Data Science at the University of Michigan; before that, a few years shipping GenAI systems in an aviation engineering org. This page is really about what's in the repos below.
- bm25-fusion — An ultra-fast BM25 implementation from scratch in Python, JIT-compiled with Numba. Supports classic BM25, BM25+, BM25L, and ATIRE variants, metadata filtering, real-time add/delete, and HDF5 persistence. It's on PyPI —
pip install bm25-fusion. - url_classification_dl — Feature extraction and engineering on raw URLs, feeding a neural classifier. My most-forked repo, so apparently it's been useful to more people than just me.
- Chaos-Game-Representation_BioSeq — Turns DNA sequences into chaos-game fractal images, then compares those images to measure similarity between genes and species. Has a Streamlit GUI and a Medium writeup if you want the full story.
- SpatioTemporalKoopman — Playing with Koopman-operator / dynamic mode decomposition methods (HODMD) for spatiotemporal systems.
- big_data_analysis — 20+ years of US flight performance data (1987–2008), crunched in Scala/Spark. A big-data course needed a genuinely big dataset.
- ACoin — A toy blockchain/cryptocurrency in Java, built to actually understand block-chaining and proof-of-work rather than just read about them.
- LlamaIndex — Triton streaming/chat inference support, and S-BERT re-ranker integration for Hugging Face's Text-Embeddings-Inference
- Hugging Face
- NVIDIA Triton Inference Server
GitHub also seems to think this adds up to a Pull Shark and Arctic Code Vault Contributor badge, for what it's worth.
- Analysis of Text-Semantics via Efficient Word Embedding Using Variational Mode Decomposition — PACLIC-35, 2021
- CT Image Enhancement Using Variational Mode Decomposition for AI-Enabled COVID Classification — 2023
- Data-Driven Volatile Cryptocurrency Price Forecasting via Variational Mode Decomposition and BiLSTM — ICICC 2022