Building AI Pipelines with Eclipse Graphene
summary
Eclipse Graphene is an incubating fork of Akumos that currently spans more than a dozen repositories and intentionally does not store binaries; the portal holds a catalog and visual design studio while deployments run in external execution environments. Modules are Docker containers exposing a protobuf gRPC interface and optional web UIs; an orchestrator generates runtime artifacts, supports gRPC streaming and parallel edges, and enables one‑click deployment to Kubernetes via the AI Playground — demonstrated with a Sudoku solver and a trainable news classifier whose model files and TensorBoard logs are saved to a shared folder.
AI-written summary · about
chapters (14)
- 0:00Eclipse Graphene intro
- 2:16System architecture
- 4:23Marketplace AI modules
- 6:36AI playground deployment
- 8:50Orchestrator communication
- 10:18Pipeline definitions deployment
- 12:30Data broker execution
- 13:59Roadmap and future developments
- 15:09Use cases for Graphene
- 17:45Demo setup marketplace
- 19:53Design studio construction
- 21:40Deploying pipelines
- 23:49Training pipeline demo
- 26:24Q&A
related talks
36mApplied AI with Eclipse Graphene
28mAI Pipelines With OPEA: Best Practices for Cloud Native ML Operati...
20mArchitecting for Innovation: Building AI Powered Apps and Platforms
40mEnd-to-end AI Marketplace with Execution Engine on Kubernetes
43mAI Pipelines With OPEA: Best Practices for Cloud Native ML Operati...
27mGen-AI at Scale: Simplifying Orchestration of Healthcare Applications Across Multi-Cluster Environme
36mNavigate the future of AI: Deploy generative models on Amazon EKS
32mDAISY - An Eclipse based IDE for Automotive Machine Learning