AI made your team faster.
I'll help make them better.
Making AI-generated code safer to ship
by leveling up the team building it.
Your agents write code fast. Nobody's sure it's right. Reviews pile up, subtle bugs slip through, and AI pull requests get a rubber-stamp 'LGTM' instead of a real look. Speed went up. But so did the tech debt.
I fix that.
Why work with me
I've spent 20 years holding the senior bar and the last year at the frontier of AI code quality.
- ✓Led, hired, and mentored engineering teams as a staff engineer, tech lead, and engineering manager.
- ✓Found and reported real security vulnerabilities in Python and Perl, with public CVEs.
- ✓Technical reviewer on ten-plus Manning software engineering books.
- ✓Taught data science and web development in Python, Ruby, and TypeScript, as an instructor at The Iron Yard.
- ✓This past year I built a production AI code reviewer, solo: roughly 400,000 lines of Go and TypeScript in about six months, and catalogued more than 6,000 documented ways AI agents get code confidently wrong. It runs 75+ deterministic checks across Go, Python, TypeScript, and Rust, and returns the same findings every run.
So I know exactly where AI breaks, and how to make it ship quality work you would put your name on. The seasoned engineer who hasn't lived inside AI tooling can't see where it breaks; the AI enthusiast often can't tell good from plausible. I've lived both.
Four ways to work with me
Four distinct engagements, from a one-time audit to ongoing work. Most start with Assess, the fixed-price audit below. Build, Coach, and Advise are longer-term and hands-on.
Assess
Find the risk.
I read your codebase by hand and with my own tooling, and I read your team the way a tech lead and engineering manager does. I sit with your engineers, watch how they actually work, and listen: how they use their tools, where their time goes, where they get stuck, and where morale, trust, or ownership is quietly costing you. You get a ranked findings report and a concrete roadmap to make AI-assisted work safer to ship. This is the fastest way to start.
Build
Build it right.
I embed as a senior engineer and architect and build your guardrails with you: custom rules that encode your standards, and gates that stop bad code before it merges. Beyond that, I help your team build AI features into your product and set up their coding agents, hands on keyboard, pairing as we go so your engineers level up and keep the skills, habits, and automated processes after I leave. Project-based or a few days a week.
Coach
Level up the team.
I level up your engineers on doing AI-assisted work well: hooks, skills, subagents, prompt technique, and the clean-code fundamentals the AI skips, SOLID, DRY, and putting correctness and security ahead of clever code. Plus the human review tools can't replace: access-control leaks, auth checks that pass when they should fail, and business logic that gives away refunds or admin rights. Done with your team, monthly or as a focused cohort.
Advise
Steer the org.
Fractional CTO and engineering-leadership help. I steer your org through the AI transition without the quality cliff: how to structure teams, set the standards and metrics that keep the bar high, decide what AI tooling to build versus buy, and hire and level up engineers for a world where the AI writes the first draft. For founders and engineering leaders who want a seasoned operator in the room.
Start with an audit.
One sprint, fixed price, starting at $5,000. I go deep on the code and the team building it, and hand you a clear, ranked picture of where you're weak and the fastest path to fix it.
- ✓A risk-ranked findings report on your real code. Security, correctness, concurrency, performance, and error handling, each with its exact location and severity.
- ✓An architecture and code-quality read: where your design will and won't scale, where structure has rotted, and the refactors that pay down the most risk.
- ✓A map of where your AI is confidently wrong, named against 6,000+ documented ways agents fail, such as tests that assert mocks instead of behavior, error handlers that swallow real bugs, and duplication dressed up as reuse.
- ✓A teardown of your AI workflow. How your hooks, skills, subagents, and rules are wired, how your engineers prompt, where reviews bottleneck, which gates are missing, and where quality slips through.
- ✓A read on how AI-assisted work actually flows through your team: where reviews bottleneck, where engineers lean too hard on the agent, and where process, ownership, or morale is quietly costing you throughput.
- ✓A 30/60/90 roadmap. What to fix now, what to guard against, what to coach.
- ✓A starter set of custom rules, hooks, skills, and subagents that encode your standards, plus a CI gate config. Yours to keep, paying off after I leave.
- ✓A live readout with your engineering leads, walking the findings and the plan.
Most teams use it to decide what to build or coach next, and many adopt Mindrealm, my code-review platform, off the back of it.
Tell me where you're stuck, in the code and on the team,
and I'll show you the fastest way to fix it.