01 · Engineering
AI-assisted development that survives production
Specifications, planning, context, evaluation, review and delivery gates. The hard part is not getting an agent to write code; it is stopping it from quietly building the wrong thing.
Hands-on training · English / Spanish
Bring a laptop.
Leave with it working.
The course assumes smart people and limited time. We build the real thing, inspect what the tools did and keep enough room for the inevitable live-demo surprise.
See the programmes ↓LanguagesES / EN
DefaultHands-on
AudienceReal teams
OutcomeIt runs
01 / Programmes
Four useful ways into the problem.
Each programme is adapted to the codebase, constraints and experience already in the room.
02 · Agents
Agents with tools, evidence and a stop button
Tool design, memory, schedules, permissions, escalation and human approval. Participants build an agent whose output can be inspected instead of merely admired.
03 · Infrastructure
Build a local AI laboratory
Model selection, quantisation, GPUs, RAG, speech, images, observability, privacy and cost. Bigger models are welcome when they earn the hardware.
04 · Web
The browser ate the desktop
WebGPU, built-in AI, Project Fugu, installability and the modern browser as an application platform. A workshop built around things participants can run in the tab.
02 / Formats
Enough theory to make the lab useful.
A focused session can establish the model. A longer engagement can change how a team works on Monday.
- 45–60 minTechnical briefing
A sharp shared model for leaders and engineering teams, followed by difficult questions.
- Half dayWorking workshop
One concrete system, built and inspected together with exercises that survive outside the slides.
- Full dayTeam lab
More build time, failure analysis and adaptation to the team’s own architecture and constraints.
- ProgrammeSeveral sessions
Practice between sessions, feedback on real work and enough repetition for a new workflow to stick.
03 / Evidence
Teaching came before the bio said “teacher.”
I started teaching cloud computing at Illinois Tech in 2010. The subjects have changed; the need to make them concrete has not.
- 2010Illinois Tech
Cloud computing and Java games; forty students with an excellent instinct for vague explanations.
- 2014—NowWeb, Polymer, PWA and browser platforms
Courses and workshops across the evolution of the modern Web.
- 2020JuntosDesdeCasa
Public workshops in programming, Blender and Unity for children and families during lockdown.
- 2025—2026Applied AI in public
Agent workshops, local AI, browser AI and a three-city Build with AI road trip.
For engineering teams