Docker Compose Your Dev Toolkit for AI & Cloud
summary
Docker Compose’s new features for AI workflows: declaratively define LLMs in Compose using a models section, access those models from application containers, connect or prepare external resources via provider services, and offload heavy GPU work remotely with Docker Offload. The talk demonstrates live examples of defining an LLM in a Compose file, using Compose to orchestrate code, models and runtime context, and handling external services and remote GPU offload for modern multi-service AI and agent architectures.
AI-written summary · about
chapters (10)
- 0:00Welcome
- 1:14AI and cloud dev challenges
- 2:59Docker Model Runner for LLM
- 10:43Docker Offload for cloud models
- 16:04MCP servers and Gateway
- 21:02Integrate Gateway in Compose
- 28:09Publish Compose apps
- 33:35Convert Compose to Kubernetes
- 38:09Services Provider for Compose
- 42:42Summary
related talks
50mCompose the Future: Building Agentic Applications, Made Simple
25mBuilding AI Workflows: From Local Experiments To Serving Users
07mKeynote: Accelerating GenAI Adoption with Containers: Unleashing AI/ML Development...
38mDesign and Implementation of Next‑Gen App Development Tools Using Compose Multiplatform × AI
35mRamaLama: Running AI Models in Containers.
46mPanel: DevX Past & Future
47mRamaLama: Making working with AI Models Boring
30mMind the Gap - A Developer's Roadmap To Building AI Agents