Learn¶
The full material — course content with the theory layer the exam actually tests, plus the professor's own revision-session walkthroughs.
The six official exam topics — where each lives¶
The end-term's official study guide names six topics. Every one is covered below — read the week section to learn it, the short note to revise it:
| Official topic | Learn it | Revise it |
|---|---|---|
| 1. Observability & Monitoring (metrics, percentiles, readiness, AI cost) | Week 6 — reading metrics | Data & ML |
| 2. Data Pipeline Integrity (identity, idempotency, provenance) | Week 6 — pipeline integrity | Data & ML |
| 3. CI/CD & Release Security (secret isolation, supply chain, rollouts) | Week 7 — release security | Git, security & practices |
| 4. Reliable AI/LLM Systems (verify output, authz in code) | Week 3 — reliability discipline | LLMs & prompting |
| 5. Web/API/Infra Fundamentals (statelessness, error design, delegated access) | Week 2 — the four new sections | Web & APIs |
| 6. Applied AI Judgment | The grading guide — it is the topic | same |
Week notes¶
Read in course order, or jump to the weeks the frequency table says matter most for you.
| Week | Topic | Time | Examinability |
|---|---|---|---|
| Bridge | Terminal, virtualenv, HTTP basics, git basics | ~8 min | 🔥 assumed knowledge |
| W1 | Git workflow, SQLite, HTTP clients, data formats, bash | ~20 min | 🔥🔥 every paper |
| W2 | FastAPI, CORS, caching, secrets, Docker, deployment | ~25 min | 🔥🔥🔥 highest density |
| W3 | Prompt anatomy, system/user prompts, Structured Outputs, embeddings | ~20 min | 🔥🔥 asked + growing |
| W4 | RAG pipeline, chunking, retrieval, grounding, stale docs | ~18 min | 🔥🔥 new — expect fresh questions |
| W5 | asyncio, agents, tools, sandboxing, MCP, memory | ~18 min | 🔥 new — GA-tested heavily |
| W6 | DuckDB/Parquet, rate limits, dedup, Playwright, ethics | ~18 min | 🔥 asked across papers |
| W7 | Prompt injection, OWASP LLM Top 10, serverless, CI/CD, Terraform | ~20 min | 🔥🔥 new — GA-tested |
| W8 | BigQuery ML, MLflow, finetuning, quantization, HuggingFace | ~15 min | 🔥 new — GA-tested |
New weeks are not less important
W4 (RAG), W5 (Agents), and W8 (ML platforms) have no PYQ history — but that's because they're new this term, not because they're less examinable. The professor said "patterns changed" and the GAs test these heavily. Expect the exam to lean on fresh content.
Revision sessions¶
The professor's own paper walkthroughs, kept as primary source:
- ET-1 — solved the T1-2026 AN paper live (27 Aug 2026)
- ET-2 — solved the T3-2025 FN paper live (28 Aug 2026) · the LLM-grading reveal
- ET-3 — no paper: the exam format itself (3 Sept 2026) · the 50% subjective split
More sessions will be appended as they happen.