◆ How I Work
I ship production AI in days because the implementation bottleneck no longer exists. TraceLayer — a compliance SaaS competing directly with Vanta — was designed, built, and live to paying customers in 21 days, solo.
Read howNo PM layer. No delegated implementation. The person on the proposal writes the code. Weekly video demos, async by default. If you have worked with agencies where you felt like a ticket in a backlog, this will feel different.
See my principlesI work with a small number of clients because the architecture decisions and production incidents that define a good engagement cannot be diluted. Your system gets my full focus.
Get in touch◆ 02 — Relationship
Most clients work with me in more than one shape over the course of our relationship. The engagement starts wherever the problem is most acute — and evolves as the company does. The relationship is the asset; the engagement mode is just how it gets billed.
Most engineering teams use AI tools the way they used Stack Overflow in 2015 — sporadically and shallowly. I work directly with your team to install the workflows and tooling discipline that turn Claude Code and agentic development into compounding leverage. Often this is the first shape an engagement takes.
HANDS-ON WORKSHOPS ◆ WORKFLOW DESIGN ◆ TOOLING SETUP ◆ 90-DAY FOLLOW-THROUGH
Once your team is operating well, the next problem is usually the system itself. I design and build production AI pipelines from the ground up — multi-agent orchestration, RAG, evaluation frameworks, MLOps — with the same discipline I used to ship TraceLayer in three weeks. Built to ship, built to scale, built to maintain after I hand it off.
ARCHITECTURE ◆ IMPLEMENTATION ◆ EVAL FRAMEWORKS ◆ HANDOFF TO YOUR TEAM
When the company is scaling and AI is core to the product, the relationship evolves into embedded leadership. Fractional architect, lead engineer, or technical advisor — inside your Slack, in your standup, owning the AI roadmap. The shape that compounds the most for both sides.
FRACTIONAL ARCHITECT ◆ LEAD ENGINEERING ◆ ADVISORY ◆ ROADMAP OWNERSHIP
The relationship usually starts with one shape. It often becomes two, then three. The way to figure out which shape fits where you are right now is a short call.
Book a call◆ Selected Work
Eight projects across compliance automation, enterprise AI, climate tech, fleet management, fintech, and open source tooling.
View allYears building
ML, LLM systems, and full-stack AI products since 2017
Systems shipped
Production deployments across IoT, fintech, compliance, and AI
Compliance frameworks
SOC 2, ISO 27001, HIPAA, PCI-DSS, GDPR, and six more
Weeks to SaaS
TraceLayer: from zero to paying customers in 21 days
◆ Writing
A precise account of the TraceLayer build: architecture decisions, the agent graph design, the integration layer, and why the implementation bottleneck is no longer the constraint it used to be.
ReadThe technical implementation is well-understood. The decision of when to use multi-agent architecture, and when a single well-prompted LLM solves the problem better, is not.
ReadHow I actually use Claude Code day-to-day as my primary development environment. Session structure, the design-first discipline, test-first as a quality gate, and where it breaks.
Read◆ Contact