Staff-level product and platform engineer
I build product systems that stay reliable when the stakes are real.
Staff-level product and platform engineer working across TypeScript, React, Node.js, AI systems, Kubernetes, and GCP. I take ambiguous, production-critical work from the first decision through rollout, measurement, and incident response — on platforms serving 24M+ users, and on my own products, where I also pick what to build, price it, and answer the support line.
Operating evidence
All case studiesEngineering work has reached users of
- Alaska Airlines
- Kitco
- ZeroHedge
- American Tower
- Staffbase
- The Wealth Advisor
- Osseus
Selected work
Systems that scale
A selection of platforms, applications, and infrastructure I've designed, built, and operated in production environments.
- Atmos Advisor — Loyalty-program training inside Team AAGTypeScript · React · Staffbase · Supabase
- InteliFactu — invoice-to-ledger automation with AI extractionTauri · Rust · Next.js · React
- Agentic infrastructure — an AI platform I built and operateClaude Code · TypeScript · Python · ChromaDB
- ZoneMirror — open-source real-time DNS sync for cPanel and CloudflarePHP 8.2 · Cloudflare API · SQLite · systemd
- Vexa Mail Insight — self-hosted DMARC observability in productionNext.js 16 · TypeScript · Drizzle · SQLite
- Denver TAQ-102 — a bargain-bin tablet on Linux, and two patches upstreamC · Linux kernel · Buildroot · DRM/KMS
- My test suite took 19 minutes. Now it takes 32 seconds.Python · pytest · pytest-xdist · cProfile
- Shared platform delivery on GKE and KubernetesKubernetes · Helm · GKE
Capabilities
What I do
I set shared delivery patterns and own production systems from architecture to deployment.
- Full-stack platform delivery: Transitioning legacy logic to modern React/Node.js stacks.
- Reliability-sensitive engineering: Scaling architectures for 30K+ global users via Kubernetes.
- Technical leadership & scoping: Translating vague business problems into architectural plans and running them end-to-end.
Technologies I work with
Open source
Code you can read
Published packages and tools with public source, CI, and releases — 10 own npm packages, with 6,475 downloads in the month to October 4, 2026.
Vexa Mail Insight
Self-hosted DMARC observability: ingests aggregate (RUA) reports over IMAP, including zipped attachments, and turns them into a queryable view of who sends mail using your domains and how SPF/DKIM perform. Next.js 16 and Drizzle over SQLite, first-run install flow, optional OIDC, published as a container anyone can pull.
syntopica/brain
The engine behind a personal LLM wiki, in the Karpathy second-brain pattern rather than RAG: page schema, ingest, link graph, query, and a lint pass that keeps the graph coherent as it grows. The engine ships public; the wiki it maintains stays private.
syntopica/clips
Turns captured web pages into wiki pages, with the author and the verifier kept deliberately apart so the model that writes a page is never the one that signs it off.
ZoneMirror
cPanel/WHM plugin that mirrors DNS Zone Editor changes into Cloudflare in real time — multi-tenant, per-user encrypted tokens, systemd-supervised worker. PHPStan level 8, 690 tests green on PHP 8.2–8.4.
eslint-plugin-code-policy
Architectural linting for TypeScript — enforces atomic files, one export per file, and explicit public APIs. Published alongside the syntopica/codeality baseline packages (tsconfig, ESLint, and Prettier configs).
syntopica/agents · BRP
Agent workflow engine that compiles one source of truth into rules and skills for ~25 AI coding IDEs (Cursor, Codex, Antigravity, Copilot…) plus a Claude Code plugin — enforcing a plan → test → review protocol with a bundled reviewer subagent.
agents-skills
Reusable Agent Skills for TypeScript, React, Next.js, Rust, and Tauri in the AgentSkills.io format — installable with npx skills add, with self-discoverable llms.txt and machine-readable skill indexes.
@favish/staffbase-cli · utils · drawer · dev
Public toolchain for building custom Staffbase apps — idempotent widget-deploy CLI, shared utilities (81-locale i18n), a Shadow-DOM-isolated drawer component, and a local dev harness. Published via npm Trusted Publishing (OIDC) and used in production for enterprise employee platforms.
Favish platform Helm charts
Public, versioned Helm charts running Drupal and Solr on Kubernetes — multi-tenant with HPA autoscaling and cron-worker decoupling, a baked Drupal Solr configset, and a backward-compatibility contract. Consumed in production across multiple high-traffic clients.
Beyond the case studies
More projects
Production systems, open-source tools, and applied engineering experiments — the wider body of work behind the selected case studies.
- ZeroHedgeClient work
- The Wealth AdvisorClient work
- Alaska Airlines ShoutoutsClient work
- Ask ATC — American TowerClient work
- Kitco platformClient work
- VerticaClient work
- Vexa Mail InsightOpen source
- 10xJoyProduct
- CaseGPTProduct
- InteliFactuProduct
- DameTicketProduct
- Tienes La VibraProduct
- MidiaProduct
- Contract signingProduct
- Public-tender intelligence platformProduct
- Invoice AgentsTooling
- POI ToolkitProduct
- Inbox CompanionTooling
- VexaPrototype
- DJ RocketProduct
- DJ Player Deluxe (Raspberry Pi)Tooling
- DMX lighting showsTooling
- LiveSalesCoach.aiPrototype
- Minted AssetsClient work
- OsseusClient work
- enlima.peClient work
- Attendize customizationTooling
- RocketUpdaterOpen source
- Server toolboxTooling
- Domain LabTooling
Working Model
How I work
I am strongest where product, backend, frontend, infrastructure, and production risk overlap. I thrive in async, remote-first environments and take ownership of what goes to production.
- End-to-end Ownership: If I build it, I run it. I monitor deployments, data pipelines, AI-assisted outputs, and user-facing behavior in production.
- Async, Written First: Technical clarity over endless meetings. I produce documentation, RFCs, and decision logs as a natural byproduct of development.
- Product and business judgment: I look for the user and business constraint behind the technical work, then shape the smallest reliable path to production.
- I design, ship, and operate my own products: a ticketing platform taking live card payments since 2021, a multi-tenant AI invoice-processing SaaS, and a DMARC observability tool released open source. Deciding what to build and living with that decision is a different skill from delivering a spec, and I practise both.
Quick info
What I'm looking for
Languages
Building a team that owns production?
I am open to remote senior and staff-level product and platform roles where technical judgment matters after the deploy, not only before it.
A useful first message
Send the role, product context, team shape, location constraints, and interview process. I will reply with the most relevant evidence from the systems I have shipped and operated.