Freelance AI work. I was building with language models and image generation before most people knew what those were, and writing about what it all meant.
Clients
Forest Contact / MobileCoin (2021-2022): Signal messenger bots with crypto wallets. I contributed to the forest framework and helped build imogen, a bot that generated images from text prompts and accepted MobileCoin payments inside Signal.
Sparkl (2022): A social app for AI image generation, DALL-E mini/Stable Diffusion era. I did frontend and design. Shipped the app in under 60 days, 10k images generated in the first month. I think we were the first webapp to allow users to remix each other’s images. Also possibly the first to use the sparkle emoji ✨ for generative AI?
Minion AI (2023): A GPT-3 agent that plays Zork, with DALL-E 2 visualizing each turn. This was before the age of agents, or even RLHF, but the base GPT-3 model could get as far as the dungeon when put in a ReAct loop.
Mori (2023): Genealogy startup R&D. Built experimental “supergenealogy” tools: an early AI research assistant with tool access to your family tree, generative ancestor simulations, living portraits.
Side projects
Memery: Search images with natural language using CLIP. 564 stars.
Silicon: Obsidian plugin that suggests related notes while you write. 129 stars, 4,400+ downloads.
Berduck: AI rubber duck bot. Created the Hellthread and other memes when it went viral on Bluesky.
Writing
The Mirror of Language mapped prompt engineering onto ceremonial magic.
When will a human level AI be built? predicted the near future of AGI.