Open-source tools I build and maintain, grouped by what they do. Each entry links to its GitHub repo and a full write-up.
Research code
Reproducibility code, a CRAN package, machine-checked proofs, and reference implementations of my research projects: Bayesian sampling, combinatorial models, random spanning forests, and deep generative models.
Combinatorial Regression — Reproducibility code for "Statistical Modeling of Combinatorial Response Data", including MH-within-Gibbs sampler, pseudo-marginal MH sampler, and HMC-NUTS sampler for Bayesian regression for responses that are integer-valued vectors subject to pre-determined constraints. Read More
combreg — R package on CRAN implementing the MH-within-Gibbs sampler from "Statistical Modeling of Combinatorial Response Data" (JASA, 2026). Read More
Anti-correlation Gaussian Data Augmentation — Code for the paper "Gibbs Sampling using Anti-correlation Gaussian Data Augmentation, with Applications to L1-ball-type Models". Read More
aldous-broder-lean — Lean 4 formalization of "The Order Mass Identity of Rooted Forests, with a Direct Proof of the Extended Aldous–Broder Theorem": the order-mass identity, the Aldous–Broder tree law, and the stopped-forest law, machine-checked with mathlib and no unfinished proofs. Read More
RecVAE — Recurrent 3D-Conv VAE for fMRI — A PyTorch reference implementation plus an 18-lesson tutorial series for modelling temporal fMRI data with deep variational autoencoders. Read More
Writing tools
LaTeX pipeline tools: clean a project into an arXiv-ready zip, and validate a UF dissertation against the formatting rules.
latex2arxiv — Submit to arXiv without the headache. CLI + Chrome extension + MCP server + GitHub Action + VS Code extension for cleaning a LaTeX project into an arXiv-ready zip, with pre-flight checks and an upload guide. Read More
latex2ufdissertation — A safety-net validator for UF doctoral dissertations. Produces a severity-tiered report citing the originating UF rule for each finding. Read More
Apps & tools
Standalone apps: a desktop app for interactive study notebooks, an academic-application tracker, and an offline quant-interview study terminal.
NoteBuilder — Local-first macOS app for interactive study notebooks: sliders, formulas, charts, and quizzes that react to each other, shared as one offline HTML file. No account, no cloud. Read More
Academic Application Tracker — Streamlit dashboard, run locally or try the live demo, that answers "what do I do today?" — tracks postdoc, PhD, faculty, and fellowship applications: deadlines, recommenders, materials, interview rounds. Read More
Quant Prep Terminal — A single-file, offline-first study terminal for quant interviews — cheatsheets, 300+ flashcards, and an in-browser options/probability lab. No install, no account, your data stays in your browser. Read More