Workshops & speaking

Your team learns by doing, not watching.

Hands-on sessions for engineering teams — from half-day deep dives to full-day installs. Each workshop is run by the founder, on your codebase, with your team. Nothing generic.

FormatOn-site or remote
Team sizeUp to 20 engineers
FacilitatorFounder-led, every time
Follow-upWritten notes + working materials
“Most AI training doesn't stick. The ones that do are the ones where the team writes actual code during the session.”

The operating model changes when everyone on the team internalizes the conventions — not just the lead who attended a talk. Workshops are how we compress that learning into a day instead of a quarter. You leave with working artifacts, not slides.

The workshops
W1
Half day · 3–4 hrs

Eval-driven development

Most teams know they should measure AI output. Almost none do it consistently. This session installs the habit: we build an eval suite on a real feature from your codebase, agree on what “good” means, and gate it in CI before we leave. You finish with a working eval harness, not a template.

On your codebaseBring a recent AI PROutput: working eval suite + CI config

What you'll leave with

  • A working eval harness on a real feature
  • Agreement on what "passing" means per task type
  • CI gate config, ready to run in your pipeline
  • A one-page "eval ownership" convention
“The eval session was the first time our team actually agreed on what a good PR looks like from an agent.// VP Engineering · Series-B fintech
W2
Half day · 3–4 hrs

AI in code review — without losing rigor

Letting agents into the PR flow creates friction unless the conventions are explicit. We work through your existing review process, name what AI can own (first-pass checks, test coverage, doc strings) and what it can't (ownership, intent, architecture), and write the conventions down. Your reviewers stop arguing about what to flag.

On your PR historyBring recent agent PRsOutput: review convention doc

What you'll leave with

  • A written AI-PR convention your whole team has seen
  • Clear ownership lines: what AI can merge, what needs a human
  • A reviewer's checklist for agent-authored code
  • One agreed-on tool for first-pass review
W3
Full day · 6–7 hrs

From prototype to production

The gap between “it works in the demo” and “it's reliable in prod” is mostly operational, not technical. This full-day session takes one AI feature your team has shipped (or is about to ship) and walks it through the entire production-readiness checklist: evals, error handling, cost instrumentation, rollback, and incident response. You leave having hardened something real.

Full day · on-site preferredBring one live featureOutput: hardened feature + runbook

What you'll leave with

  • One production-hardened AI feature
  • Cost + latency instrumentation in place
  • Incident runbook, tuned to your setup
  • A production-readiness checklist you can reuse
“We'd been ‘almost production-ready’ for two months. The full-day session got us actually there in six hours.// Engineering Manager · SaaS platform
W4
Full day · 6–7 hrs

Agent design for engineering teams

Not every task benefits from an agent. This session teaches your team to think like a system designer: we map the tasks in your backlog, score them on autonomy-readiness, design the first pipeline as a group, and build a working prototype before we leave. Your team finishes knowing where to invest next quarter — and having proved one of those bets.

Full day · on-site preferredBring your task backlogOutput: working pipeline prototype

What you'll leave with

  • A scored autonomy-readiness map of your backlog
  • One working pipeline prototype
  • An agreed "agent design checklist"
  • A roadmap of where to invest next
“One day of focused work with the whole team is worth six months of hoping the practices spread on their own.// CTO · Developer-tools company
Custom workshops
Built around your problem

Don't see the session you need? Let's design it.

The four standard workshops cover the most common gaps. If your team has a specific problem — a particular pipeline, a codebase with unusual constraints, a mixed-seniority group that needs different things — we'll scope a bespoke session. Same format, same founder-led delivery, same working artifacts at the end.

Book a workshop

Tell us about your team.

The founder reads every inbound and will reply within two business days with availability and a short pre-call to understand your team's setup.

Which workshop(s) are you interested in?
// The founder reads every inbound. Expect a reply within two business days. No automated responses.