Projects
Things I've Built
Open-source research tools, high-performance libraries, and infrastructure. Most of my research software is written in C/C++ for performance, with Python bindings where appropriate.
dir2text
Personal
Python library and command-line tool, published on PyPI, that renders directory trees and file contents in LLM-ready formats.
LPmade
University of Notre Dame
A comprehensive link prediction framework for complex networks. Provides tools for generating features, building models, and evaluating link prediction methods with proper methodology.
VCP
University of Notre Dame
Vertex Collocation Profiles: software for subgraph counting and link analysis. Implements the VCP framework for characterizing local network structure around node pairs.
mRMR
McGill University
Substantially improved C++ implementation of the minimum Redundancy Maximum Relevance feature selection algorithm by Peng et al. High-performance implementation with Python bindings.
PostgreSQL Genomics
McGill University
Relational database representation for genotypic data. Efficient storage and retrieval of genomic variant data using PostgreSQL user-defined types implemented in C.
WEKA SMOTE
University of Notre Dame
Author of the SMOTE (Synthetic Minority Over-sampling Technique) component for the WEKA data mining platform. One of the most widely used oversampling methods for imbalanced classification.
kdtree
Personal
A fast C++ k-d tree for nearest-neighbor search.
popc
Transamerica
A C++20 implementation of the POPC clustering algorithm, with Python bindings.
Home Lab Infrastructure
Personal (private repositories)
28 self-hosted services with separate DEV, QA, and PROD environments. Kubernetes (k3s) with Flux GitOps, Authelia and LDAP single sign-on, OpenBao secrets, Harbor with Trivy scanning, Renovate, Gitea CI, and shared LLM, embedding, and speech gateways.