Skills

A practical map of what I can actually help with.

My strongest work sits where product judgment and implementation meet: shaping the right thing, building it carefully, and debugging the real boundaries between code, services, users, and releases.

Product engineering

25+ years turning ambiguous product ideas into shippable interfaces and workflows.
Balancing user experience, implementation cost, and release timing.
Engineering leadership, architecture, product delivery, and team development across startups, enterprise, healthcare, and consulting work.

Web applications

React, Next.js, Astro, Tailwind CSS, and component architecture.
Dashboard, editor, onboarding, and account-management interfaces.
Frontend state, routing, rendering boundaries, and performance-aware UI work.

Mobile applications

React Native, Expo, Android Kotlin, iOS, Xamarin, and .NET MAUI experience.
EAS build and submit workflows, native configuration, and release readiness.
Production mobile work across subscriptions, diagnostics, camera/media controls, biometrics, SaMD applications, and app-store delivery.

Auth, billing, and data

Clerk, Supabase auth, organization context, and user-account boundaries.
Stripe and RevenueCat integration work across web and mobile products.
Convex, Supabase, PostgreSQL, pgvector, SQL Server, background jobs, and email workflows.

AI and agent workflows

Claude Code, Codex, OpenCode, MCP, Vercel AI SDK, Gemini Flash and Nano, ML Kit, and prompt/context management.
LLM-assisted product features, semantic search, embeddings, retrieval workflows, and browser/productivity tooling.
Agent workflows where model output stays inspectable, reviewable, and connected to real product work.

Delivery and debugging

Tracing failures through real configuration, deployment, and provider boundaries.
Release hardening, migration paths, storage access, and operational cleanup.
CI/CD, GitHub Actions, Azure DevOps, Cloudflare, Vercel, Datadog, Sentry, and production observability.

Operating style

Grounded, direct, and implementation-aware.

I prefer to inspect the real system before making confident claims. That usually means reading the repo, checking the provider contract, tracing the deployment path, and then making the smallest useful change.

The recent agent and LLM work fits the same pattern: useful automation, visible outputs, and human review loops that keep the product owner in control.