Interactive study command centre

Your prep,
at a glance.

Twenty-two hands-on labs with 282 simulations you can operate and break on purpose, and the 304 topics behind them: Kubernetes, GCP, LLMs, RAG, agents, multi-tenancy, architecture and leadership.

0topics studied

0day streak

0categories done

0starred

⊞ Revision cheatsheet 🔥 Take the hot seat

Study activity

Three tracks. They are independent (start whichever suits you) but the order within each is deliberate, and each lab still stands on its own if you would rather not follow it. New to all of it? Open Cloud lab for foundations, or ADK lab to build something today. Revising? The cheatsheet puts every lab’s must-remember points on one printable page.

Build, run and govern an agent: 10 labs, in order

The main track, and the one most people want. Build one, run it in production, hand the building to a coding agent, measure it, secure it, publish it, then the capstone carries a single agent through every stage at once.

1 ADK lab Build a working agent, give it tools, and understand why it picks the ones it does. 10 tabs · 11 simulations · first agent to production routing

2 Pick your frame Build the same agent eight ways, see what each framework's opinion costs, and untangle the Lang-family names. 10 tabs · 14 simulations · eight frames, one loop

3 AgentCore lab Take an agent that works on your laptop and make it one you would trust in production. 9 tabs · 11 simulations · scenario drill

4 Agents building agents Write a specification precise enough that a coding agent builds the right thing from it. 9 tabs · 10 simulations · spec to running service

5 Evaluation & hill climbing Answer “did that change make it better?” with a number instead of an opinion. 11 tabs · 14 simulations · metrics to unit economics

6 The number to trust Pick the metric that fits the mistake, and defend a threshold with named stakes. 11 tabs · 14 simulations · one class of twenty throughout

7 Govern & secure agents Decide what an agent is allowed to do, and stop it doing everything else. 12 tabs · 14 simulations · gateway to audit trail

8 Gemini Enterprise rollout Put an agent in front of a whole company without breaking identity, search or trust. 12 tabs · 15 simulations · identity to rollout

9 Peak Week, one agent end to end Carry one agent through all six stages in a single sitting, and see where each one bites. 10 tabs · 16 simulations · groundwork → work → survive → safe → good → theirs

10 The FDE bootcamp, day one: build an enterprise agent Take an agent apart into its organs, wire five specialists on three runtimes together, then attack it and watch which hook stops the leak. 8 tabs · 15 simulations · anatomy → protocols → identity

Cloud and architecture: 5 labs, in order

How the cloud actually works, what each provider calls the same jobs, how to draw it so somebody else can act on it, and then the two exam workbenches. Useful on its own, and the ground the agent track is standing on.

1 Cloud lab Follow one request from a phone to a database and name every box it passes through. 14 tabs · 27 diagrams · 30 simulations · 94 services mapped

2 The three-cloud atlas Type a service name from any cloud and read across to the other two, small print included. 7 tabs · 7 simulations · 786 services · 175 jobs · three clouds

3 Architecture diagrams Draw a diagram that survives a review and tells people what to do. 11 tabs · 11 simulations · a synthesis you may argue with

4 GCP Architect exam Read a case study the way the exam does, and find the clause that rules the others out. case workbench · 4 simulations

5 SAP-C02 cheatsheet Hear a phrase in a question and know immediately what it is really asking for. trigger drill · decision tables

Under the hood: 6 labs, in order

The layers everything above is running on. Optional depth rather than a prerequisite, take it whenever you want to know why inference is priced the way it is, and read it in order.