Learning path
Agents in Production
Cost and latency, caching, deployment and monitoring, with animated films and runnable Python. Part of the Agentic AI Engineer premium track.
-
1
Cost and Latency
Make agents cheaper and faster: measure cost per step, route easy steps to smaller models, run independent tool calls in parallel, cap output length, cut steps, and stream…
-
2
Caching
Three caches every agent needs: prompt caching for long repeated prefixes (system prompts, tools, documents), tool-result caching with expiry times, and response caching, plus how to keep caches…
-
3
Deployment and Monitoring
Run agents reliably for real users: background jobs with a queue and status endpoint for long tasks, idempotent jobs, the metrics that matter (success rate, p95 latency, cost,…