01Lumen
A computational photography lab — an exposure-triangle simulator rendered live on real photographs. Drag aperture, shutter and ISO; watch depth of field, motion blur and grain respond like a real camera.
Try the live demo →Independent engineering studio
Websites, web apps, and AI infrastructure for startups and small teams — designed, engineered, and delivered by a senior engineer, without the agency overhead.
Selected work
Engineered end to end — robust by design, stable in production, novel where it counts.
01A computational photography lab — an exposure-triangle simulator rendered live on real photographs. Drag aperture, shutter and ISO; watch depth of field, motion blur and grain respond like a real camera.
Try the live demo →LLM inference cost estimator — model size, quantization, GPU type and traffic in; GPU count, latency and the monthly bill out. The back-of-envelope math behind every serious serving deployment.
Try the estimator →Concept site for a fictional coastal restaurant — editorial typography, an interactive menu, and a reservation flow. Built with Astro, zero stock photos.
Personal finance dashboard — budgeting, spend tracking, and trends. Built end to end as an exercise in shipping complete, maintainable products.
Interactive tutor for learning music theory — exercises, instant feedback, and progress tracking, all in the browser.
Webhook automation for small teams — route events between the tools you already use, with retries, run logs and zero YAML. A boring-problems-done-well SaaS concept.
Services
Three lanes, one standard: robust, production-grade systems — delivered on time.
01
Marketing sites, web apps, and storefronts that load fast and convert. Modern stack, clean code any developer can pick up.
02
Put LLMs into production: inference serving, RAG pipelines, and GPU performance tuning — the unglamorous work that makes AI products actually work at scale.
03
Dashboards, integrations, and workflow automation that give a small team leverage. Small builds, outsized returns.
Process
No account managers, no black boxes. You work directly with the person writing the code.
A short call to understand what you're building, what success looks like, and whether we're a fit. No pitch decks.
Fixed-scope proposal with a timeline and a price. You know exactly what you're getting before we start.
Weekly demos and async updates. You see progress early and often — no big-bang reveals.
Clean code, docs, and a walkthrough. You own everything — and I'm a call away if you need me after.
About
Shipwright Studio is an independent engineering studio run by a senior engineer with 8+ years building networked systems at cloud scale — from datacenter networking to the GPU communication infrastructure behind large-model training and inference.
I started the studio to do the work I do best: hard problems, robust systems, software that ships and stays up. You work directly with the person writing the code — no handoffs, no telephone game.
Based in the San Francisco Bay Area, working with clients worldwide. MS in Electrical & Computer Engineering, UC Davis.
Contact
Tell me about your project — I reply within one business day.