Cato, Zürich. I design, ship, and operate full production systems. Fractional CTO / principal engineer.

Six live, CI-green, test-covered web platforms — three of them past 2,300 commits — running on infrastructure I operate myself.

OrangeCat — Bitcoin-native economic platform. Live pay-link and commerce surfaces, Lightning settlement. Exchange, fund, lend, invest and govern are the roadmap, not what ships today. 2,774 commits · 2,648 tests · CodeQL + Playwright E2E + uptime checks.

evig — storefront, community marketplace, repair and IT-help, workshops, AI-assisted intake for refurbished hardware. 2,630 commits · 7,783 tests · 151 migrations · 8-job CI.

FleetCrown — execution OS and AI-agent orchestration: hosted control plane plus a local daemon that owns shell, git and PTY to drive coding agents. Pre-1.0, live. 2,345 commits · 9 workflows · signed desktop binaries across 13 releases.

Hirnli — Swiss fundraising intelligence: 16,900 foundations turned into actionable grant applications. Multi-tenant SaaS. 907 commits · 950 tests.

Kivvi — AI-first ERP for Swiss SMEs and secondhand retail. Monorepo, Dockerised. 650 commits · 1,316 tests.

AOZ Wohnen — compatibility-based housing placement: 38 factors, conflict prediction, full audit trail. Deployed demo with sample data. 3,748 tests across 215 files — the highest test-to-code ratio I run.

Swiss civic & public-service — Reparaturbonus Zürich (repair-bonus platform, residents to certified repair shops) · SBB Lost & Found (real-time lost-item recovery, Redis pub/sub + Socket.io) · revamp-info (IT support knowledge base)

Clinical & care portals — VitaReBa (metabolic psychiatry and longevity: booking, care teams, GDPR-scoped clinician AI) · Surf Your Life (burnout and Long COVID recovery — daily check-ins, symptom tracking, early practitioner intervention)

Commerce & booking — Petvity (pet care platform) · S.Ink (studio portfolio and booking) · printcraft (scene composer for physical art)

AI & data infrastructure — DataCat (forms, photos, documents and audio in; validated records and dashboards out, with multi-provider vision failover) · Botsmann (domain-specialised AI professionals) · Solon (on-chain treasury, signed votes, decisions tracked against KPIs)

• One verify command per repo — lint, typecheck, test. CI runs exactly that. Green locally means green on the branch.
• Self-hosted by default. Hetzner, Caddy, systemd, Postgres. No vendor lock-in, no per-seat pricing, no surprise bill.
• Everything ships through a branch and CI. The deliberate step is the pipeline, not a person clicking a button.
• One automation harness across the whole fleet — golden CI templates and audits that run from fleet against every repo.
• Built to hand over. Documentation, backup restores, and runbooks included.
• TypeScript · Next.js · Postgres · Drizzle/Prisma · Tailwind · GitHub Actions

Each one extracted from a system above, published tokenless via OIDC trusted publishing.

• @bitbaum/ai-kit — stay on free LLM tiers: multi-vendor fallback, 429 classification that tells the three kinds apart, fair-share rationing of a shared daily pool.
• ai-forms — fill a form from one sentence, then keep talking to it. The hard part isn't the first fill, it's that "actually, make it urgent" has to patch the form instead of wiping it.
• threadkit — headless multi-participant threads. Permission is participation, unread is per-person, and an AI participant obeys the same visibility rules as everyone else.

Rates and scope: bitbaum.github.io/hire

• Fractional CTO — own the architecture, the deploy pipeline and the engineering bar for a team that doesn't have a senior engineer yet.
• AI-native product builds — greenfield, fixed scope, delivered in weeks.
• Agent fleets & internal automation — the recurring internal work that eats a team's week.
• Rescue work — an inherited codebase with no tests, no deploys, and nobody left who wrote it.

Zürich, Switzerland · orangecat.ch