LET'S BUILD
AI THAT STAYS.
Your AI roadmap exists. Your team is capable. Months in, nothing has shipped at scale. I get it into production that lasts, without burning the people who run it.
A billion events a day at Yammer. Nextdoor through an IPO. a16z Speedrun SR005. Kubernetes contributor.
Founders I have backed
"Glo helped us sharpen our pitch and our strategy before we raised. Six months later we raised our seed from 500 Global."
"We walked in needing fundraising help. Glo tore our pitch apart, rebuilt it, and made the intros. We got into Character Labs."
Three things I do
Pilot to production that stays
I get AI past the demo and into the systems your customers actually touch, built to survive the next year, not the next sprint.
- Kill the pilots that will never ship
- Pick the one path to real users
- Evals, guardrails and rollback before launch
- Your team owns the on-call, not me
Adoption your team keeps
Tools stick when the people who maintain them helped shape them. Speed that does not cost you your engineers.
- Maintainers in the room from day one
- The hard thing said early and in private
- Skills transferred, not dependencies created
- A team that keeps shipping after I leave
Judgment that has been there
I made the expensive mistakes already, on someone else's clock. You get the conclusions without the tuition.
- Yammer to a billion events a day
- Nextdoor through an IPO
- a16z Speedrun SR005
- Kubernetes contributor. First Terraform provider for Oracle Cloud.
I work with a few teams at a time, at the principal level. If you need a contractor to burn down a ticket backlog, I am the wrong call. If your AI strategy is one demo away from a board meeting and you cannot afford for it to be theater, we should talk.
I have spent fifteen years shipping systems that survive growth. Yammer to a billion events a day. Nextdoor through an IPO. My own company into a16z Speedrun.
I have also watched AI hype torch good teams: a roadmap turned into a graveyard of proofs-of-concept, and engineers who flinch at the word. So I build the durable kind. AI that reaches production, stays there, and leaves your team stronger than it found them.
- Adoption is a people problem wearing a technical costume.
- Speed that burns the team is not speed. It is debt with a deadline.
- I say the hard thing early, in private, with care. That is most of the job.
- The goal is a team that keeps shipping after I leave, not one that depends on me.
What happens when you click
A 30-minute call
You tell me what you are building and where it is stuck. I tell you what I see. No pitch.
A scoped engagement
We pick the one thing that matters most and I go at it with your team, on a defined scope.
On-call or retained
I stay for as long as the work earns it, and I leave your team able to run it without me.
Questions, answered straight
Are you available, or running your own company?+
On call. I take a few engagements at a time. LTNT is a product I build. It is proof I still ship, not my day job.
Do you write code, or just advise?+
Both. Kubernetes contributor, author of the first Terraform provider for Oracle Cloud. I am in the codebase when that is where the problem lives.
Is this coaching?+
No. This is engineering leadership for teams adopting AI at scale. Founder coaching still exists. It is a separate door.
How do engagements work?+
A 30-minute conversation where you tell me the problem. Then a scoped engagement. Then on-call or retained, as the work needs.
LET'S BUILD AI
THAT STAYS.
Tell me what you are building and where it is stuck. I will tell you what I see, and whether I am the right person to help.