Day 5. Picked observability because both AWS apps and LLM apps are basically opaque without it, and most tutorials hand-wave it.
Topics to cover:
- CloudWatch Logs + Logs Insights — structured logging and useful queries
- CloudWatch Metrics + alarms — custom metrics, dashboards, what's worth alerting on
- AWS X-Ray — distributed tracing, service map, sampling
- LangSmith / OpenTelemetry for agents — token, tool-call, and prompt traces
- Cost and latency observability for LLM apps — what to track per request
Plan: Chandana on the AWS-native stack (CloudWatch, X-Ray), me on the agent-side stack (LangSmith, otel), and a shared note on what to do when an agent calls AWS and you need both traces stitched together.
Day 5. Picked observability because both AWS apps and LLM apps are basically opaque without it, and most tutorials hand-wave it.
Topics to cover:
Plan: Chandana on the AWS-native stack (CloudWatch, X-Ray), me on the agent-side stack (LangSmith, otel), and a shared note on what to do when an agent calls AWS and you need both traces stitched together.