Sociotechnical Platform Engineering
summary
Argues pipelines are explicit value streams and delivers a reusable value‑stream map plus Wardley-style dependency maps that map dev → UAT → prod and where error budgets enforced stop feature work. Shows tooling and patterns: medallion data architecture, Databricks/Snowflake style data lakes, production Python notebooks, MLOps steps like vector embeddings and RAG and model evals, plus feature‑flagged deploys and timing diagrams. Calls out operational constraints: SLIs→SLOs→SLAs alignment, cloud/GPU cost risk and when to shift to chaos‑testing.
AI-written summary · about
chapters (8)
- 0:00Introduction and speaker background
- 1:31Pipelines as value streams
- 3:15Defining DevOps and SLOs
- 5:43Worldly maps for process steps
- 9:16Aligning SLOs with DevOps practices
- 13:10Data platforms and MLOps
- 15:40Continuous delivery with a rubber chicken
- 16:41Platform engineering and closing
related talks
25mSociotechnical Platform Engineering
28mThe DevOps Pipeline
41mPlatform Engineering for Developers, Architects & the Rest of Us
27mOptimizing the engineering value stream
41mOrganizational Sustainability with Platform Engineering
24mBringing Product Thinking into DevOps
45mPlatform Engineering: A Deep Dive Conversation
33mBuilding Resilient (ML) Pipelines for MLOps