Hendrik Louw

I make engineering teams faster with AI. Hands on keyboard, not slides: I'm a staff engineer running agentic development in production every day, with the guardrails and eval gates that make it safe.

Why AI pilots stall

MIT's 2025 study of enterprise AI pilots found that 95% produce zero measurable return. The models weren't the problem. Those teams had no harness around their agents and no eval gate deciding what merged, so nothing ever earned the trust of the senior engineers who had to live with the output. The pilots stayed pilots.

I close that gap from inside the codebase. I lead teams, and I ship with AI every day: agentic coding loops inside guardrails, with eval gates and review deciding what merges.

Track record

Five years leading engineering at a UK telehealth scale-up

Online clinical assessment, prescribing and subscriptions, in regulated healthcare. I owned the platform end to end through five years of category growth, including medically supported weight loss. The engagement renewed monthly, 60-plus times.

Client name and detailed case study on request.

AI-driven bio-data platform for a US food-safety biotech

My current engagement: platform engineering for pathogen-intelligence products. I'd never worked in the scientific domain and was shipping useful commits in week one. The harness did the onboarding.

Detailed case study on request.

How we can work together

Team workshop: AI-assisted engineering that ships

2 days · hands-on in your codebase

Agent loops, guardrail patterns, eval gates and review policy, worked through in your own repo. Your team leaves with a rollout plan your senior engineers helped shape, which is the only kind they adopt.

AI engineering audit

2 weeks · fixed scope, fixed fee

Two weeks inside your team's real workflow: how you build with AI today, where the guardrails are missing, and what to fix first. You get a written plan and an executive readout.

Fractional AI engineering lead

1-2 days/week · 3-month minimum · 2-3 clients at a time

Embedded in your team. I install the tooling and the metrics, train your engineers, and ship the hard parts myself. The alternative is a $400k head-of-AI hire.

What I believe about AI-assisted development

"The agent wrote it" is never an excuse. The harness is what makes the speed safe.
Adoption is a leadership problem before it's a tooling problem. Senior engineers adopt what protects their standards and quietly ignore what doesn't.
The metric that matters is shipped scope per engineer with quality intact. Everything else is theatre.

Start a conversation

If your AI ambition is ahead of your shipped results, that's the conversation I'm useful in. Tell me what you're trying to ship and where it's stuck.

hendrik@hendriklouw.dev

or book directly: calendly.com/hendrik-hendriklouw/30min