GitOps Meets AIOps: Building Reliable AI Pipelines With Argo
summary
Argo orchestrates LLM agents to find and patch open‑source CVEs end‑to‑end, using Argo Workflows for per‑agent DAGs, Argo CD for GitOps deployment, and Carpenter for just‑in‑time node scaling; agents post results to GitLab MRs and Jira and persist artifacts to S3. Pipelines enforce clear input/output contracts, retries, Prometheus/Grafana metrics, and human approval gates. A surprising config limits web scraping to 80,000 hits per agent. Current gaps: database rate limits and shared agent context (now MD files, moving toward vector DB).
AI-written summary · about
chapters (14)
- 0:00AI in production pipelines
- 1:29Speakers and JFrog platform
- 3:18Agentic vulnerability remediation
- 4:02Agent AI versus predefined rules
- 5:30Security remediation challenges
- 6:34AI agents fixing CVEs
- 8:38Scaling fixes with orchestration
- 9:45Argo workflows fit
- 12:10AI pipeline infrastructure
- 13:35Argo workflow layer breakdown
- 17:07Workflows separation of concerns
- 19:33Monitoring AI vs traditional workflows
- 20:36When to use AI and when to avoid it
- 21:17Current challenges and improvements
related talks
43mGitOps at Scale with and Argo CD on
26mGitOps Me Some of That! Managing Hundreds of Clusters with Argo CD
13mHow GitOps Should I Be? Lessons from a Kinda-GitOps Environment
33mGitOps Pipelines: Everything Everywhere All at Once
24mEmpowering the GitOps Future with AI
05mLessons from a Large-Scale GitOps Implementation
27mImplementing GitOps with ArgoCD: A Practical Guide
44mAgentic Delivery, Guardrailed: AI in CI/CD & Platform Engineering (Open Source Patterns)