All three paths share the same foundation, then specialise. Read the first lesson of each to see which one fits you.
BUILD AI APPLICATIONS
AI Engineer
19 modules · 54 lessons · 13 labs · Free
Build products that use LLMs: grounded answers, structured outputs, evaluation and deployment.
For you if
You like building backends and products, and want to ship AI features that hold up in production.
You’ll be able to
- Build an authenticated AI API with validated inputs, deadlines, and controlled failures.
- Create a versioned ingestion and retrieval pipeline with citations and tenant isolation.
- Compare models and prompts using held-out cases, release gates, latency, and cost.
What you’ll build
A deployed knowledge assistant that answers from documents with citations, an evaluation report and a rollback plan.
BUILD AGENTS
Agentic AI
31 modules · 86 lessons · 27 labs · Free
Build tool-using agents with memory, planning, durable execution and human approval.
For you if
You enjoy systems design and want to build AI that takes steps — safely and within limits.
You’ll be able to
- Implement a bounded agent loop with validated tools and explicit termination.
- Persist and recover tasks without duplicating side effects.
- Connect scoped tool servers and build evidence-backed research workflows.
What you’ll build
A support agent that gathers evidence, proposes actions for approval, recovers from crashes and is tested on its full trace.
DELIVER FOR CUSTOMERS
Forward Deployed Engineer
31 modules · 84 lessons · 25 labs · Free
Take AI into real companies: discovery, integration, pilots, rollout and handoff.
For you if
You like working with people and code, and want to own whether AI actually works for a customer.
You’ll be able to
- Translate a customer workflow into a scoped pilot with a measurable baseline.
- Design and implement data and API integrations that respect identity and permissions.
- Run customer acceptance sessions and manage defects against agreed gates.
What you’ll build
A customer pilot from discovery to handoff, with a data readiness report, acceptance tests and a defended case study.