01Flagship · Case study
trajecktory
Running a job search the way a revenue leader runs a pipeline.
A local career command center that runs my entire job search from one dashboard. It scans job boards, scores each role against my resume, and tailors a resume and cover letter per posting. It schedules the follow-ups so nothing goes cold, works the right contacts through a recruiter and in-network CRM, and drafts LinkedIn and X posts in my voice. Every AI-written message is mine to edit before it sends.
One placethe whole job search, sourcing to offer
hrs→minscan, score & tailor a resume per posting
Your voiceAI drafts every message, you approve the send
Built with Claude CodeNode + ReactSolo buildOpen source
02Live arena · Case study
Silicon Showdown
An arena where AI agents play games on real hardware.
A public arena where AI agents play twelve strategy and party games on declared hardware while people watch every move and private thought. Agents, models, GPUs and quantizations each get a leaderboard. I designed it, specified and reviewed every line with Claude as orchestrator and Codex as builder, and ran it like a revenue operation: route the work, test the measuring instrument, cap the cost before the spike.
~400matches in the first 8 hours
3.83→4.30clarity moved by one prompt sentence
84test suites, re-run on every change
Live at siliconshowdown.comCloudflare WorkersClaude + CodexAgent evaluation
03SDR tool · Open source
Outreach Grader
Grade your cold outreach. Know exactly what’s broken before you hit send.
Most cold outreach fails for the same handful of reasons — and you can’t see them from the inside. Paste a cold email, LinkedIn DM, or connection note, say who it’s going to, and it scores the message across eight dimensions, quotes the exact words holding it back, and rewrites the whole thing on request. The rubric encodes years of SDR leadership: what a 3 looks like versus an 8, and why. Runs locally through Claude Code — no API key, no account.
Built with Claude CodeRuns locally8-dimension rubricNo API keyOpen source
04Library · MIT
ai-text-hygiene
Extracted from trajecktory and released on its own.
npm: ai-text-hygiene
A small npm library that strips the invisible characters riding along in AI-written text: zero-width spaces, smart quotes, and watermark-style marks. Extracted from trajecktory, released MIT.
MITnpm: ai-text-hygieneExtracted from trajecktory
