Matthew Billman's Portfolio.
This is a collection of my open-source code.
I make music too: matthewbillman.com
A Python 3.14 library + Typer CLI that scrapes the PyCon US 2026 schedule, lets you search it three ways (Levenshtein-fuzzy, exact-keyword, or Voyage AI semantic embeddings), build a personal saved-list, and one-way sync it to Google Calendar via OAuth.
Highlights: async + on-disk-cached HTTP, stackable --day / --from / --to / --room
filters, inline negative search (!python, !"machine learning"), conflict
markers, idempotent gcal sync with drift detection, and every magic default
configurable via config set. 137 tests, 83.8% coverage, ruff + ty clean.
A collection of rust applications I’m particularly proud of.
A python library for AI pair programming, centered in the principles of Test Driven Development. Developed for research purposes (whitepaper), though may eventually be fully developed as a pytest plugin (time permitting).

DSPy is a framework designed to optimize language model (LM) prompts and weights algorithmically, particularly in complex pipelines. It simplifies the traditionally messy process of manually tweaking prompts and finetuning steps by separating program flow from parameters and introducing new optimizers.
Various adaptions of the Transformer machine learning architecture, for use in varied NLP and NLP-adjacent fields:
A personal mental health dashboard developed in Flutter + Dart.
A COVID testing and patient monitoring system developed in partnership with the Camai Community Health Center, the State of Alaska, and Cue Health.
An e-commerce, blog, student portal, and performance diagnostics site build for ibidPREP in Manhattan, NY.
A test grading and report generation software that serves as the glue between ibidPrep’s many sources of data.
Quick configuration of Traefik, Docker Compose, and FastAPI for HTTPS webapps.