AI coding tools work great on toy projects. Greenfield experiments, demos, clean repos. That's not the hard part. The hard part is everything else. Production environments. Legacy architecture. A codebase that's been running since before the AI tools existed.

This is for you if:

The tools fall apart on your real codebase, not a demo
You've tried it seriously and can't figure out why everyone else seems to be getting results you're not
You're expected to lead your team on this and you're not sure you believe in it yourself
You're hoping this blows over. You know that's not a strategy.

Who I am

I use Claude Code as my daily driver on a mature, production Ruby on Rails codebase. After being the VP of Engineering at Academia.edu for 4 years, I stepped back to IC.

I have been at Academia.edu almost 12 years and have done every job in engineering there is - full-stack product, SRE, infra lead, director, VP, and now IC again. Being a mentor to engineers has been a constant throughout.

Before Academia.edu: I was an independent software vendor in the late 90s, worked at Goldman Sachs, co-founded a startup that got into YCombinator, got fired from a well-known tech company in 2009, and ran a consultancy for six years. It took me 15 years to stop pinballing around.

How this works
Three steps. Each is a prerequisite for the next.

Step 1 - AI Era Career Strategy (start here)
A single 60-minute session. We talk about your codebase, your role, what you're worried about. You leave knowing what your actual problem is and what to do about it.
Sometimes the answer is Step 2. Sometimes it's something you can fix on your own and I'll tell you that too.
$150 / 60 minutes

Step 2 - AI-Native on Legacy Code (the core work)
Three sessions, two hours each. $1,800, sold as a package.
You bring a real ticket from your real codebase. We work it together.

Session 1: you watch. I drive.
Session 2: you drive. I correct in real time.
Session 3: you drive. I am mostly quiet.

I designed this to make itself obsolete. After three sessions you don't need me.
$1,800 / 3 sessions. A real ticket on a real legacy codebase is very much preferred. I will happily sign an NDA. If these are not an option, we will figure out how to make it work for your specific case.

Step 3 - Continued Guidance (for graduates only)
For engineers who have finished the package and want an ongoing thought partner. We meet twice a month. You bring whatever you're stuck on. Use it to stay current as the tools keep changing, or just to have someone to gut-check decisions with.
This plan works best if you have completed the AI-Native on Legacy Code package. Exceptions possible - let's discuss.

If this sounds familiar, start with Step 1.