Advisory
For teams already running production inference and paying a real GPU bill. I take the system apart, break down what is actually happening, and sit on the calls that move cost and latency. You keep the implementation.
I take only 2 clients a month. Not a hire. Not a Joule service.
Start here
$5,000 USD
A focused live review and written findings memo, scoped to 5 specific questions or traces you name.
One week. 100% credited toward Month 1 if a retainer starts within 30 days.
Base scope up to 5 traces. Complex multi-cluster setups scoped and billed separately.
Then the seat
$10,000 / month
A standing seat on the serving stack: compute economics, agentic systems, or both.
Three-month minimum. Scoped to 2 concurrent engineering workstreams.
Secures your seat. We kick off within 5 business days. If intake shows an irreconcilable scope mismatch, the payment is refunded immediately.
A decade building production ML. Author of LLMOps (O'Reilly). GPU Engineering: AI Inference and System Design (Packt) is almost finished, launch is end of 2026 or early 2027. Invited Faculty for AI Inference Engineering at Andreessen Horowitz Academy (The Academy SF).
Checkout
Payment unlocks the calendar and the intake. I take only 2 clients a month.