01
AI agents & LLM products
From copilots and voice interfaces to autonomous workflows — LLM-powered features grounded in your data, wrapped in production infrastructure, evaluated the way real products are evaluated.
Division 01Technology & AI
Engineered for the next decade of work.
AI-powered products, custom software, and the cloud architecture underneath. This is where the ecosystem's leverage begins — systems that turn human intent into repeatable outcomes.
Most organizations don't have a technology problem — they have an execution surface problem. Requirements outrun the tools; teams ship faster than the ground beneath them can hold. My work sits in that gap: designing the systems, choosing the right primitives, and shipping software that keeps compounding after I leave the room. Every engagement starts with the same question — where is leverage hiding, and what is the smallest, most durable thing we can build to unlock it?
What I build
Four practices, one engineering standard.
02
Custom software & SaaS
Full-stack builds when the off-the-shelf tool doesn't fit. Type-safe TypeScript, React, and edge-native backends — designed to be shipped in weeks and maintained by the team you already have.
03
Cloud & infrastructure
Right-sized architecture on Cloudflare, AWS, or Supabase. Serverless-first where it earns its keep, boring where boring wins. Observability, cost, and security modeled from day one.
04
Integrations & data pipelines
The unglamorous connective tissue: enterprise integrations, event pipelines, ETL, and internal APIs that finally let your systems talk to each other.
How I work
A build cadence engineered for velocity without regret.
- 01
Diagnose the real system
One week embedded with your team. Map the workflows, the workarounds, and the friction the roadmap keeps missing. Deliver a written brief with three specific bets.
- 02
Prototype the smallest bet
Ship a working prototype of the highest-leverage bet in 10–14 days. Real code, real data, no slides. If the bet is wrong, we learn cheaply.
- 03
Harden into production
Type-safety, tests, monitoring, cost model, incident runbooks. The prototype becomes a system your team can run without me on call.
- 04
Hand off with leverage
Documentation, architecture decision records, and a training arc so your engineers own the codebase — not a black box built by a consultant.
Selected work
Recent engagements.
AI · Product
Agentic workflow platform
A private LLM-powered ops layer for a services firm — replaced 40+ hours a week of manual triage with a supervised agent loop.
SaaS · B2B
Multi-tenant credentialing app
Full-stack SaaS on TanStack Start + Supabase, launched in 9 weeks with SSO, audit logs, and a public API on day one.
Infra · Migration
Edge-first re-platform
Moved a Node monolith to Cloudflare Workers with typed RPC and per-request tenancy — 6× cost reduction, sub-100ms global TTFB.
Next step
Have a system worth building right?
If you're weighing an AI product, a rebuild, or a hard integration, start with a 30-minute conversation. I'll tell you honestly whether I'm the right build partner.