Workshop

Loop Engineering: Reliable Work From Coding Agents

Stefan Jansen

CEO, Applied AI; Author, ML for Trading

Reliable work from coding agents starts with the loop, not the prompt.

Coding agents do their best work inside a loop you can trust: a spec that kills ambiguity, durable context, bounded work units, a verification gate, and a handoff that survives the session. In one focused afternoon you'll build that loop around a real repo — and meet the failure mode that matters most: an agent that keeps everything green while quietly doing the wrong thing. You leave with files that run and a pattern you reuse — not notes, not a recording.

If you saw the Lightning Lesson, you watched me catch the bug. The workshop is where you build the machinery — the spec, the gates, the skill — in your own hands, on a repo you keep.

What you’ll learn

You'll leave with a working loop and the files that prove it: a spec, a plan, a shipped change, a verification skill, and a handoff.

  • Use an interrogation pattern that forces ambiguity out before any code is written.

  • Leave with a spec.md that states "done" as conditions a check can confirm.

  • Project the plan onto GitHub issues so state lives outside the chat window.

  • Ship one issue end-to-end — branch, PR, and a verify gate — on a real repo.

  • Hit a planted trap where every automated check passes but the result is corrupted.

  • See why an agent optimizing for green takes the cheapest path your checks allow.

  • Encode the invariant — not the example — for your own domain, live with me.

  • Add the judgment layer most agent demos skip: knowing what to verify, and how.

  • Write a handoff a fresh session — or a different agent — picks up cleanly after a /clear.

  • Apply cost discipline and the build order: manual → skill → loop → schedule.

Workshop agenda

  • Frame: prompting vs. loop engineering

    Why most people cap out at 20%, what changes when you stop being the engine, and the honest “do you even need a loop?” test.

  • Scope it: turn ambiguity into a checkable spec

    An interrogation pattern that pins down what “done” means — as conditions a check can confirm, not a paragraph of hope.

  • Project onto durable work units

    Plan → issue → branch → PR. The loop’s memory and boundaries live outside the chat, so nothing is lost between sessions.

  • The gate: green-but-wrong (the core)

    Hit a planted trap where the suite stays green while the result is corrupted, then write the verification skill that catches it. Plus maker≠checker and cross-agent review — Claude writes, Codex reviews.

  • Survive the boundary

    Write a handoff a fresh session or a different agent can actually use, apply cost discipline, and learn the build order: manual → skill → loop → schedule.

Learn directly from Stefan

Stefan Jansen

Stefan Jansen

Contact

Stefan Jansen is the author of Machine Learning for Trading (3rd ed.), maintainer of six open-source Python libraries and 450+ teaching notebooks, and founder of Applied AI. For the last year-plus he has built production systems with coding agents across Claude Code and Codex — the exact loop-engineering discipline this workshop teaches.

See all products from Stefan

Who this workshop is for

  • You use a coding agent on real, sustained work — and you've lost sessions, re-explained context, or shipped work you couldn't fully trust.

What's included

Stefan Jansen

Live sessions

Learn directly from Stefan Jansen in a real-time, interactive format.

Lifetime access

Go back to course content and recordings whenever you need to.

Community of peers

Stay accountable and share insights with like-minded professionals.

Certificate of completion

Share your new skills with your employer or on LinkedIn.

Maven Guarantee

Your purchase is backed by the Maven Guarantee.

Frequently asked questions

Maven for Teams

Reimbursement

Get your company to pay

Everything L&D needs: email template, receipts, and certificate of completion.

Get reimbursed

Team discount

Learn with your teammates

Save 20%+ when 2 or more teammates enroll in the same cohort.

Save 20%+ with a team

Private cohort

Run a cohort for your org

A dedicated cohort with a custom schedule and curriculum, tailored to your team.

Book a private cohort

$400

USD

Jul 18
·

10am–2pm EDT

Enroll