Programmes
Learn to build AI agents and make stronger architecture decisions.
Live, project-based Agentic AI programmes with Sunil Mathew
For experienced professionals who want to implement AI agents, review their design or guide the team building them. Develop practical skills and engineering judgment through one reliable agent, a system of several agents or the shared Architecting Agentic AI Systems learning project.
These programmes examine what an agent may do, how to evaluate its behaviour and where human review belongs.
The questions every production agent must answer
What may the agent do? What evidence should it have before acting? How will you know a task completed? What happens when the next attempt repeats work that already succeeded?
Bring those questions into a scoped project. Build, test, discuss the design and use feedback to revise it.
Choose your programme
| Programme | Who this is for | Learning outcome | Hours | Fee | Action |
|---|---|---|---|---|---|
| Architecting Agentic AI Systems | Engineering leaders, architects and staff engineers who need to build, review or guide AI-agent work. | Build and evaluate a shared support-agent learning project through deployment, including control, recovery, evidence and architecture review. | 30 live, 6 weeks | $1,399 | Request a fit call |
| Design and Build Multi-Agent Systems | Senior engineers, technical leads and architects whose problem needs more than one agent. | Build a working multi-agent system, explain its architecture and evaluate where it holds and where it breaks. | 20 live, 5 weeks | $999 | Request a fit call |
| Build Reliable AI Agents | Developers, product and programme managers exploring one focused workflow; hands-on scope depends on technical starting point. | Build a bounded AI-agent workflow and explain its tools, behaviour, human review points and limitations. | 10 live, 3 weeks | $499 | Request a fit call |
Fees are per participant. Independent work is additional: 4–5 hours a week for Build Reliable AI Agents and 3–4 for Design and Build Multi-Agent Systems. For Architecting Agentic AI Systems, confirm the full setup and pre-session preparation commitment before joining.
Every programme runs as a live cohort of 8, never more than 10. The next Architecting Agentic AI Systems cohort starts online on Saturday 31 October 2026 at 10:00 IST and runs for six weeks; the other two programmes start once 8 participants have confirmed. Sessions are not recorded. Each participant receives a certificate of completion from Sunil Mathew; it is not accredited by a university or certification body.
Work through deployment and operating decisions
Choose Architecting Agentic AI Systems to develop the shared support agent through six stages, including deployment in an agreed learning environment. Engineering leaders, architects and staff engineers examine reliability, recovery, authority and cost alongside the build.
Design a system of several agents
Choose Design and Build Multi-Agent Systems when one agent is not enough. Design the roles and handoffs between agents, build the system and evaluate where it holds and where it breaks. Suitable for senior engineers, technical leads and architects.
Start with one focused workflow
Choose Build Reliable AI Agents when you have a bounded use case and want to understand how its responsibilities, tools and context connect. Developers, product managers and programme managers can discuss a scope suited to their technical starting point.
Which one fits?
| Programme | Pick this if | Best for | What you get |
|---|---|---|---|
| Architecting Agentic AI Systems | You will design, review or approve AI-agent systems and need to defend those decisions. | Engineering leaders, architects and staff engineers | The judgment to choose an architecture, set what agents may do, decide release readiness and plan rollback, practised on a full system |
| Design and Build Multi-Agent Systems | Your problem needs several agents working together. | Senior engineers, technical leads and architects | A working multi-agent system with designed roles, handoffs and evaluation |
| Build Reliable AI Agents | You are building your first serious agent. | Developers, and product or programme managers exploring one workflow | A working agent with controlled tools, human review and tested behaviour |
For the engineer building it and the leader reviewing it
Engineers practise implementing and evaluating a workflow: start with Build Reliable AI Agents, or Design and Build Multi-Agent Systems when the problem needs several agents.
Architects and staff engineers examine the choices across the system: Design and Build Multi-Agent Systems or Architecting Agentic AI Systems.
Engineering managers and directors develop the questions needed to guide a team through a review: Architecting Agentic AI Systems.
Product and programme managers can start with Build Reliable AI Agents, with the hands-on scope matched to their starting point.
Your starting capability and project goal determine fit. The live hours sit alongside independent work; scope and schedule are discussed before joining.
Learn with an engineering practitioner
Sunil Mathew: 26 years in engineering, with past roles at Google, Amazon and Walmart Labs.
The programmes connect a working result with the reasoning needed to defend its design. Read Sunil's background.
Bringing a work-related use case?
Yes. Describe it at a high level, and we agree a version that is safe to build in the learning environment. Taking it into your organisation's systems needs its own engineering and operating review.
For a whole group, explore enterprise learning. An employer can also sponsor an individual place.