Forward Deployed

Production AI Systems: sections and modules

T4Required

Production AI Systems

Run AI systems safely at cost and scale.

For whom: All technical paths
Time: 29 h 4 min
Before this track: LLM and Retrieval Fundamentals, Agentic Systems

Objectives

Write an eval plan, SLOs, a security model and a cost model for one system

Evaluation and security are the most common follow-ups.

T4.1
Enterprise RAG in production
3 h 53 min · 7 lessons
7 lessons
T4.2
Evaluation and release gates
3 h 57 min · 6 lessons
6 lessons
T4.4
Security, tenancy and prompt injection
4 h 5 min · 6 lessons
6 lessons

Exercise: Eval plan for a P2 system
Done when: Checkpoint ≥ 70%

Cost, latency and observability decide production viability.

T4.3
Observability and production diagnosis
3 h 7 min · 6 lessons
6 lessons
T4.5
Cost and latency optimisation
2 h 59 min · 8 lessons
8 lessons
T4.6
Scaling and reliability
2 h 56 min · 6 lessons
6 lessons
T4.7
AgentOps: versioning, rollout, CI/CD, red teaming
3 h 4 min · 9 lessons
9 lessons
T4.8
Model adaptation and inference serving
5 h 3 min · 4 lessons
Partly ready
4 lessons

Exercise: P4 Productionize the RAG prototype
Done when: P4 rubric ≥ 3