Jason Derr
GTM engineer at Magnetiz. I build AI systems for B2B revenue teams, and every one shows its work.
This profile was put together from Jason's public GitHub portfolio, and they haven't joined Vibetown yet, so reach them through their portfolio. Is this you? Contact us to update or remove it.
About
Jason Derr works in AI GTM & Growth and AI engineering, based in California (magnetiz). GTM engineer at Magnetiz. I build AI systems for B2B revenue teams, and every one shows its work. Public projects on GitHub include System Map (Derive a cited system.md from an agent-built repo and reconcile the declared design against the code,…), Landed (A scheduled reconciler that joins what an agent CLAIMED against what actually LANDED) and Job Radar (A job-search radar for Claude Code that scrapes fresh postings, rejects the ones your rules kill (quoting…), mostly written in JavaScript, Python and TypeScript.
Projects 6
- github.com
System Map
Derive a cited system.md from an agent-built repo and reconcile the declared design against the code, deterministically. Every line cites file:line or is marked Unknown. Zero dependencies, 428 tests. ADR drafting from the diff is phase 2.
- github.com
Landed
A scheduled reconciler that joins what an agent CLAIMED against what actually LANDED. Three states only: matched | contradicted | unresolved. Fail-closed, zero dependencies.
Job Radar
A job-search radar for Claude Code that scrapes fresh postings, rejects the ones your rules kill (quoting the line), scores the rest and hands you a ranked queue. No model decides.
github.comSignal Desk
A signal-to-outreach pipeline where every decision can be inspected, replayed, or refused. Eight fail-closed stages, draft-hash-bound human approvals, sealed tamper-evident ledgers, keyless demo. Zero dependencies, 848 tests.
github.com- github.com
Business Brain
Ask a question about the business, get a one-page visual answer. A probabilistic generator wrapped in a deterministic grounding gate: every fact must quote canon verbatim (shown on the page), plus a lexical + LLM faithfulness layer. redaction-gate and gtm-agent-evals run inside it.
- github.com
Account Scout
A GTM account-research agent whose claims must survive: every claim binds to a citation fetched in the same run. The system above the primitives — webhook-engine, redaction-gate, gtm-agent-evals, and gtm-architect run inside it.