Course

Production AI Systems

Operate probabilistic systems with SLOs, evals, and fallbacks.

IntermediatePlatform, SRE, and architecture teams taking AI past prototype.Coming soon — self-paced, cohort, or hybrid

Outcomes

Define eval gates that actually block bad releases.
Trace model, tool, and retrieval hops as one distributed system.
Design gateways for routing, spend, and provider failure.
Apply queues, retries, and idempotency to agent side effects.

Who It Is For

Platform engineers
SREs
Architects owning production AI

Prerequisites

From Prototype to Production workshop topics, or equivalent production experience

Curriculum

Evals and observability

Offline/online evals
Tracing
Cost telemetry

Gateways

Routing
Rate limits
Fallbacks

Reliability

SLOs
Idempotency
Incident handling

Distributed architecture

Queues
Durable workflows
Failure domains

Architecture Labs

Labs focus on architecture decisions, not toy coding exercises.

Design an eval gate
Map an incident from prompt to tool

Included Resources

RunbooksChecklistsLabs

Instructor

Hammad Abbasi — AI & Security Architect. 16+ years across identity, distributed systems, and production agent architecture. Speaker at EIC KuppingerCole.