GenAIOps - OpenSearch for AI & GenAI Platform Observability
summary
End-to-end GenAIOps pipeline using OpenTelemetry -> OpenSearch -> Jaeger collects metrics, logs, traces and baggage from Kubernetes-hosted LLMs, Vector DBs, agents, RAG and inference pipelines and stores them for dashboards and alerts. Instrumentation captures token counts and embedding drift, model name, latency and cost; recommendations include custom spans and baggage headers and strict context propagation. Prometheus/Grafana handle metrics, Langfuse records prompt traces, and eBPF/Tetragon add network-level signals; the deployed setup proves workable for prompt tracing, cost analysis and drift detection.
AI-written summary · about
chapters (8)
- 0:00GenAIOps & OpenSearch intro
- 2:35Observability in GenAI era
- 5:27GenAI components & data flow
- 9:25Monitoring GenAI stack layers
- 12:21Critical GenAI metrics
- 13:44MLOps vs GenOps
- 15:31GenAI observability best practices
- 18:01GenAIOps summary & future
related talks
27mAlert Fatigue to Action: Observability as the Backbone of Practical AIOps with OpenSearch
42mAIOps for Distributed Environments - Deep Dive
29mNon-Invasive AI Agent Observability With OBI
22mFrom Telemetry to Insight: Building AI-Powered Observability Pipelines with OpenSearch
06mSponsored Keynote: OpenSearch - See Everything: Open Observability for Agentic AI
27mObservability as a foundation for AI-enabled platforms
25mFrom Logs to Metrics: OpenSearch's Prometheus support with Agentic AI and...
29mOpen Source Sovereign AI: Building Enterprise GenAI Platforms for Control & Comp...