I architect, ship,and operate products
16 products.One operator.
Each one shipped, deployed, and operated as production software — not coursework or proofs of concept. Three were forced offline by regulators. Six are deprecated but archived as reference. The current active product lines are the Swift and Tap suites at the top of the grid.
xtractmp3
Video → MP3 / MP4 from YouTube, TikTok, Instagram, X, Reddit, SoundCloud + 20 more sources. Self-hosted Cobalt + tikwm fallback, Cloudflare D1 + Workers, Stripe billing, OAuth, ad-free Pro tier.
Keep what you love.
Everything I'llreach for in production.
Picked deliberately — boring infra where it counts, fast tools where it doesn't. Every line in this list has shipped to a real domain with real users.
- TypeScriptDefault. Every web product.
- JavaScriptGame loops, Canvas, Node services.
- PythonApolloAPI, FastAPI services, PDF gen.
- SQLPostgres-first, RLS, transactional schema.
- GLSLPost-FX shaders for VFX Studio.
- BashCutover playbooks, infra scripts.
- React · Next.jsMost consumer + SaaS surfaces.
- Three.js · R3FVFX Studio · Atlas · Apex Akina v0.
- Canvas 2DApex Akina pseudo-3D + drift physics.
- CesiumJSAtlas surface-detail geospatial mode.
- Node · ExpressProvably-fair engines, APIs.
- FastAPISimDex market services.
- React Native · ExpoKineo PWA + native flows.
- Vite · WebpackBuild pipelines, lazy chunks.
- Cloudflare Workers · D1 · KVEdge deploy, security headers.
- Supabase · PostgresAuth, RLS, transactional schema.
- StripeSubscriptions, one-off, webhooks.
- Render · Vercel · RailwayMulti-service deploy + cutover.
- Custom DNS · TLS · domainsProduction cutover playbook.
- CI / CD (GitHub Actions)Build / test / deploy pipelines.
- System & API designREST, webhooks, auth flows.
- Performance engineeringProfiling, bundle, render perf.
- Provably-fair / cryptoHMAC-SHA256 commit / reveal.
- Real-time / WebSocketsLive chat, betting feeds.
- Cutover opsApex domain swap with zero downtime.
- Release storytellingChangelogs, launch posts, copy.
Six rules thatearned themselves.
None of these started as opinions — each came from something breaking in production and getting rewritten.
Ship to a real domain or it didn't happen.
Localhost demos and Figma flows lie. Real users on a real domain at real production cost is the only evidence that survives review.
Own the whole loop — code, infra, story.
Hand-offs are where products die. The same person who wrote the SQL writes the changelog, picks the DNS provider, and answers the support email.
Boring infrastructure, fast iteration.
Postgres over experimental DBs. Stripe over rolling-our-own billing. The novelty budget is for the product, not the plumbing.
Operate what you build.
Observability, rate limiting, security headers, and rotation procedures aren't day-2 work — they're the actual job. A live product is a system, not a milestone.
Cut scope before cutting quality.
The features that survive launch are the ones polished to the point a stranger trusts them. Everything else is a fork in the road, not a requirement.
Resilience is reading the regulator first.
Three of my products were shut down by governments. The right move is to plan exits, document handoffs, and keep your customer trust above your novelty stack.
Three phases.Architect → Ship → Operate.
The same loop every product runs through — whether it's a weekend tool or a paid SaaS. The boring discipline is what makes the velocity look unreasonable.
Cut the surface to its smallest viable shape.
- 01Map the user journey to surfaces, not features
- 02Pick the stack that fits the failure modes — not the trend
- 03Sketch the schema before any UI
- 04Identify the cutover plan from day one
Real domain, real Stripe, real auth — day one.
- 01Scaffold with the production stack — no demoware
- 02Wire payments + auth + email before the first feature
- 03Push to apex with security headers + rate limits in place
- 04Write the launch post in parallel with the last commit
Watch metrics, rotate secrets, ship the changelog.
- 01Logs + observability from day one, not day fifty
- 02Quarterly secret rotation + dependency audit
- 03Customer-facing changelog every two weeks
- 04Document the shutdown story before you need it
From kernel driverto landing page.
The app, the infrastructure, the release pipeline, and the story — one integrated system, owned top to bottom. No hand-offs.
Solo on every product.
AI-aware engineering for the productsmost teams won't finish.
I'm a product engineer based on the West Coast, solo on every product in the showcase above. Six years in, the pattern is consistent: pick a real problem, ship the smallest surface that solves it to a real domain, then operate it until the answer is "keep going" or "cut scope."
Background in WebGL, real-time systems, payments, and cutover-ops. Three of the products in this portfolio were shut down by regulators — same engineer, same skill stack, different category outcome. Resilience is part of the work, not a footnote.
Available for product / frontend / release-system work — contract, fractional, or a clean handoff of a single surface.
Got a product
that needs to ship?
One email. I'll reply with what's feasible inside the timebox you have, and what I'd cut.
No DMs, no forms, no "hop on a call." Email-first keeps the bar real and lets you check the work before we sync.