My professional mission is to bring into the world malleable software: computing environments where anyone can mold their tools to their own unique needs.

I'm currently a senior researcher at the independent research lab Ink & Switch, leading the malleable software research track.

Previously, I completed a PhD at MIT CSAIL in the Software Design Group advised by Daniel Jackson. Before that, I spent five years as an early engineer and designer building the edtech startup Panorama Education (YC S13).

These days, I'm thinking a lot about AI + malleable software: How can LLMs help people make custom software?

If you'd like to hear about my future work, you can subscribe to occasional email updates. I'm also active on Twitter and Mastodon and have an RSS feed.

Projects

Riffle

Building data-centric apps with a reactive relational database

Potluck

Dynamic text documents for creating personal software

Peritext

A CRDT for rich text, to enable asynchronous collaborative writing

Wildcard

Customize a web app with a spreadsheet! (Onward 2020)

Cambria

Better cross-app data compatibility using bidirectional lenses

Margin Notes

Automatic code docs w/ recorded examples from runtime (LIVE 2018)

Experiments

TodoMVC Vis

Visualizing the internal behavior of TodoMVC in Redux

JSON Sheets

A reactive programming environment for building GUIs

ladybug

A Ruby debugger backend for the Chrome Dev Tools UI

Experiments in Dynamicland

Exploring in the dynamic medium

Writing

Codifying a ChatGPT workflow into a malleable GUI
Malleable software in the age of LLMs
ChatGPT as muse, not oracle
Fuzzy API composition: querying NBA stats with GPT-3 + Statmuse + Langchain
For your next side project, make a browser extension
Dynamic documents // LLMs + end-user programming
Bring Your Own Client
How tweet threads cured my writer's block: Twitter as a medium for sketching
Foam: Software as Curation
Climate art, hacking Autodesk, and other funemployment projects
Browser extensions are underrated: the promise of hackable software
Adding tail call optimization to a Lisp interpreter in Ruby
ENHANCE!: Upscaling images CSI-style with generative adversarial neural networks
No Redis Needed: building a Postgres-backed recommendation engine for Rails
The Virtual Cafe: a better way to hang out remotely
Building Waze for the Boston subway: my first adventure in civic hacking
2014: A year in review with iPhone pedometer data

Contact me

Please reach out if you're interested in chatting about ideas, or if I might be able to help you in some way. You can contact me via email or on Twitter.

📦 The Archive... (older work less relevant to my current interests)

Conference Talks

ENHANCE! (BangBangCon 2017)

Explaining Generative Adversarial Neural Networks using CSI

Ruby, A Family History (RailsConf 2018)

Exploring how Ruby was influenced by Lisp, Smalltalk, and Perl

Deep Learning: an Introduction (BostonRB)

An introduction to deep learning techniques for Ruby developers

Other software projects

classroulette

Making online course catalogs skimmable and serendipitous

simple_recommender

User recommendations for Rails apps, using Postgres

Postgres BSON

Extending PostgreSQL with faster JSON queries

Data visualization / NLP

Panorama Free Response Analytics

Using NLP to help educators understand survey feedback

News context sidebar

Automatically providing background context about news articles

Pedometer data analysis

Analyzing a year of data from my smartphone step counter

Hardware / Electrical Engineering

Autonomous quadcopter

A sonar-based obstacle avoidance system for a quadcopter

Laser scanning

An algorithm for efficiently determining object orientation with a cheap laser scanner

Bulldogs Racing

Embedded software for a gas-electric hybrid racecar