Meet Asya🎭 - a new open-source Kubernetes-native async actor framework for orchestrating AI/ML workloads at scale.

GitHub repo: https://github.com/deliveryhero/asya ⭐

Image generated with OpenAI

Documentation Structure¶

Getting Started¶

• Motivation - Why Asya exists, problems it solves, when to use it
• Core Concepts - Actors, envelopes, sidecars, runtime, and system components

Architecture¶

• Architecture Overview - Deep dive into system design and components
• Actors - Stateless workloads with message-based communication
• Sidecar - Message routing and transport management
• Runtime - User code execution environment
• Operator - Kubernetes CRD controller
• Gateway - Optional MCP HTTP API
• Crew - System actors for flow maintenance
• Autoscaling - KEDA integration details
• Protocols - Communication protocols between components
• Transports - Message queue implementations

Installation¶

• AWS EKS - Production deployment on AWS
• Local Kind - Local development cluster
• Helm Charts - Chart configuration reference

Quickstart¶

• For Data Scientists - Build and deploy your first actor
• For Platform Engineers - Deploy and manage Asya infrastructure

Operations¶

• Monitoring - Observability and metrics
• Troubleshooting - Common issues and solutions
• Upgrades - Version upgrade procedures

Quick Links¶

• GitHub Repository
• Examples
• Contributing Guide