Embedding AI: Don’t Burn Your Fingers in the Heat of the AI revolution
summary
Claim: don’t treat AI as a single API call — build a platform so models run reliably and legally. Practical tools and outcomes: Kubernetes/OpenShift for autoscaling GPUs, container registries and SBOMs for supply‑chain safety, GitOps and observability for rollout and debugging, llama‑stack and VLM/LLMD for routing and standardized LLM deployment, local LLMs or RAG for frequently changing FAQs while avoiding external API data leaks. Demonstrations include an AI‑generated slide and local LLM proof‑of‑concept that work; fine‑tuning proves costly—fine‑tuning costs (tens of millions).
AI-written summary · about
chapters (9)
- 0:00Introduction
- 1:47Build your own AI platform
- 9:13AI techniques and trade-offs
- 12:05Embedding AI in enterprise
- 16:46Modern app delivery
- 26:55Uncontrolled AI risks
- 34:26Data flows and model deployment
- 39:06Monitoring AI workloads
- 42:01Q&A
related talks
36mAvoiding 5 missteps that undermine your AI readiness and success
40mOpening Keynote: AI & Platform Teams
24mAI Governance for developers: A practical framework for deploying autonomous coding agents safely…
37mThe AI Factor: The Blueprint for Intelligent Deployment
33mEngineering AI Applications: A Conversation with AI Leaders
33mOperationalizing GenAI Applications with DevOps Practices
08mshares insights on AI and regulatory compliance
14mFull-Stack AI Engineering from AI-Native Coding to Knowledge, Models, Infrastructure & OSS