44 things I've built, with what each does. API cost is what the AI coding-agent work on each would have cost at the API prices in effect when it ran; output tokens are what the agents wrote.
Sort by
Show
Benchmarks for the Bend language against OpenMP on CPU and CUDA on GPU.
A cheap transcription service: upload audio or video and get a transcript from NVIDIA's Parakeet model, with free daily transcripts and an $8 unlimited plan.
github reputation to surface quality human contributions.
Sycophancy Bench: measures whether AI models change their answers under evidence-free pushback, compared with how they respond to real corrections.
A sourced encyclopedia of UFO cases, people and claims, built as a static site.
Matching decompilation for x86-64 C++: checks candidate source against the target binary and says what to change so the compiled bytes match.
One CLI for running bounded, unattended jobs across coding agents (Codex, Claude Code, Cursor, Devin, OpenCode and more), with isolation, deadlines, resumable sessions and recorded results.
Token-usage analytics for AI coding tools: reads local Claude Code, Codex, Gemini and other session logs and breaks usage down per project with DuckDB.
LLM API prices today or at any past date, so past usage is priced at the rates that applied when it ran.
Chooses which agent takes each delegated task, by expected cost per successful task and room left in its subscription, and keeps a ledger of how every task went.
A native desktop client for Signal: one program of about 14 MB with no browser inside.
Agent terminal and multiplexer where harnesses are first class.
Routes messages between AI coding agents in Wakterm and Telegram or Signal, so agents can talk to each other and to my phone.
A fast Zsh distribution with autosuggestions, history search, syntax highlighting and directory jumping built in, measured faster than the usual plugins.
Searches past Claude Code, Codex and other coding-agent sessions by regex.
A small native Twitch app for Linux and Windows: see which channels you follow are live and watch one, with no embedded browser.
A native YouTube Music player with a music visualizer, written in C on gesso: one 4 MB executable using about 26 MB of RAM.
A fast static site generator in a single binary.
Local speech recognition and synthesis on one ggml build: Parakeet with Silero voice detection to hear, and a voice model to speak, for talking to coding agents.
OpenAI's Codex CLI with loops, completion timestamps, tool-free sessions and same-turn failure handlers added.
A chat that never ends: a summary tree of the whole history, with tools to open the original messages
A small C library on SDL3 for lean native apps and games that build for Windows, macOS, Linux and the web, with installs measured in megabytes.
Plans the cheapest week of meals that meets your nutrition targets, from real grocery prices.
Plan and quota of every coding-agent account, from Claude Code and Codex to Antigravity, Cursor and Z.ai, behind one set of commands, a statusline and a daemon.
Where each coding-agent harness keeps its home and session history, and which tools read it.
Harvest: Massive Encounter (2008) decompiled back to matching C++ and ported to macOS, Linux, Windows and the web.
Crimsonland (2003) decompiled back to matching C/C++ that plays in the browser, plus a Python port that matches it tick by tick.
A lean Markdown viewer and editor for Windows, macOS and Linux, built on gesso.
Score video encodes against their source on the GPU: VMAF, banding, lasting texture and grain, every frame
A compact, agent-friendly summary of GitHub activity since your last check.
Field reports from letting AI agents run research campaigns on real projects, and what they got right and wrong.
Autonomous trading research for a market-making competition, with agents proposing and testing strategies against held-out market data.
Shared GitHub Actions that publish all my packages, so the release steps change in one place.
A coding font: D2Coding with Nerd Font icons and emoji, the font this site uses.
Autonomous Research Bench (ARB)
Autonomous Research Bench, a proposed benchmark that scores the AI research system rather than its answer, by how fast it finds competitive solutions.
Compares U.S. economic results under Democratic and Republican presidents, with permutation tests and false-discovery correction.
Uploads videos to your own YouTube channel from the command line, with captions, thumbnails and resumable uploads.
A text-first forum with verified eligibility and anonymous posting: the browser proves membership with BBS signatures without revealing the X account behind a post.
A browser extension that improves GitHub's interface.
A compact Git-aware Zsh prompt with its own asynchronous Git collector.
A drone flight-control stack: train a policy in a fast simulator, then fly it in the VelociDrone simulator through a real controller path.
A turn-based tactics board game on a hex grid for 2 to 8 players, with a Rust engine in the browser, AI opponents and online play.
AlphaZero-style training for m,n,k games like tic-tac-toe in Rust, 75 times faster at self-play than a reference implementation.
Tracks CPU prices against benchmark scores to show which processors are good value.
At the Bank of Canada, I worked across trading, portfolio management, quantitative modelling, and policy before co-leading a $20 billion non-USD credit portfolio. At DELV, Euler, and Veda, I built Python trading and fuzzing infrastructure, worked on market strategy, and analyzed protocol risk using on-chain and internal data. Full resume