Learning roadmap
Choose a goal, then follow each stage as an ordered, explained outline.
You have not recorded evidence yet. Start with the first recommended topic, then mark the evidence you can demonstrate.
Your progress is available for this visit only because browser storage is unavailable. Keep this page open or record your evidence elsewhere.
Build a durable mental model, then transfer it into a working application.
Stage purpose: Establish the vocabulary and first working mental model.
Required background: Complete the linked prerequisites below or bring equivalent experience.
Expected outcome: Explain the core idea and use it in a small example.
Estimated time: 70 minutes
Prerequisites: No prerequisite topics
Evidence: Practise the outcome in code and record what you learned.
Current status: Not started
Recommended next: Secure Financial Write Operations — Transfer the safe-write contract to a higher-consequence financial example and compare the additional controls.
Prerequisites: Safe Write Operations in Agentic Systems
Evidence: Practise the outcome in code and record what you learned.
Current status: Not started
Recommended next: Human in the Loop for Agentic Write Operations — Add a human decision boundary before the write so high-risk actions cannot execute autonomously.
Stage purpose: Connect the concepts into a repeatable problem-solving model.
Required background: Complete the linked prerequisites below or bring equivalent experience.
Expected outcome: Apply the model to a representative exercise without copying a recipe.
Estimated time: 35 minutes
Prerequisites: Secure Financial Write Operations
Evidence: Practise the outcome in code and record what you learned.
Current status: Not started
Recommended next: Persistent Agents & Context Management — Follow suspended execution into persistent history, rewind, and recovery across long-running agent threads.
Stage purpose: Transfer the concepts into a realistic implementation task.
Required background: Complete the linked prerequisites below or bring equivalent experience.
Expected outcome: Produce a small artifact that proves the decision works.
Estimated time: 75 minutes
Prerequisites: Human in the Loop for Agentic Write Operations
Evidence: Practise the outcome in code and record what you learned.
Current status: Not started
Recommended next: Securing Agentic AI Systems — Complete the production boundary with ingress and egress defenses against adversarial instructions and sensitive data leakage.
Prerequisites: Persistent Agents & Context Management
Evidence: Practise the outcome in code and record what you learned.
Current status: Not started
Recommended next: Complete this path by applying these outcomes in your own project.
Prioritise high-frequency material, retrieve it quickly, and practise explaining it clearly.
Stage purpose: Secure the concepts that recur in most interviews.
Required background: Complete the linked prerequisites below or bring equivalent experience.
Expected outcome: Give a concise, accurate explanation with one concrete example.
Estimated time: 24 minutes
Prerequisites: No prerequisite topics
Evidence: Answer aloud, give a concrete example, and identify the relevant trade-off.
Current status: Weak
Recommended next: Secure Financial Write Operations — Contrast reversible and financial writes, then justify how irreversibility changes the control surface.
Prerequisites: Safe Write Operations in Agentic Systems
Evidence: Answer aloud, give a concrete example, and identify the relevant trade-off.
Current status: Weak
Recommended next: Human in the Loop for Agentic Write Operations — Move from mechanical write safety to the placement and lifecycle of a human approval gate.
Stage purpose: Rehearse implementation choices that invite common follow-up questions.
Required background: Complete the linked prerequisites below or bring equivalent experience.
Expected outcome: Walk through a realistic API decision and defend the trade-offs.
Estimated time: 12 minutes
Prerequisites: Secure Financial Write Operations
Evidence: Answer aloud, give a concrete example, and identify the relevant trade-off.
Current status: Weak
Recommended next: Persistent Agents & Context Management — Connect approval checkpoints to persistent state, forensic inspection, and time-travel debugging.
Stage purpose: Prepare the details that distinguish a deeper answer.
Required background: Complete the linked prerequisites below or bring equivalent experience.
Expected outcome: Connect the concept to reliability, complexity, or production behavior.
Estimated time: 27 minutes
Prerequisites: Human in the Loop for Agentic Write Operations
Evidence: Answer aloud, give a concrete example, and identify the relevant trade-off.
Current status: Weak
Recommended next: Securing Agentic AI Systems — Finish with the security sandwich and explain why prompts alone cannot enforce a trust boundary.
Prerequisites: Persistent Agents & Context Management
Evidence: Answer aloud, give a concrete example, and identify the relevant trade-off.
Current status: Weak
Recommended next: Complete this path by applying these outcomes in your own project.