distributed-workers

Orchestrate distributed workers by deploying a central ServiceOrchestrator that dispatches operations to heterogeneous workers (Backtest, Training, Host Services) across containers and native hosts. The pattern provides progress tracking, cancellation support, and GPU-first routing to reliably execute and monitor scalable workloads.

Implement workers using the WorkerAPIBase pattern, which provides automatic operation tracking, health endpoints, and push-based self-registration with the backend orchestrator. This abstraction handles async operations, progress bridges, and cancellation lifecycle endpoints out of the box.

Yes. FastAPI integrates with the distributed-workers pattern through the ServiceOrchestrator and WorkerAPIBase abstractions, supporting environment-based configuration, remote queryability, and async operation handling across microservices and heterogeneous compute environments.

Create a new worker type by extending the WorkerAPIBase pattern to inherit automatic operation tracking and self-registration. Implement your analysis logic, define health and progress endpoints, and the worker will integrate seamlessly with the central orchestrator for distributed execution.

The pattern provides built-in progress bridges and operation lifecycle endpoints that enable GPU-first routing decisions and cancellation handling. Configure workers with environment-based settings and the orchestrator automatically manages resource allocation and cancellation propagation.

Use distributed workers for scalable, computationally intensive tasks like backtesting and training where heterogeneous processing needs—GPUs, native hosts, containers—must coordinate reliably. The pattern excels when you need progress tracking, cancellation support, and independent worker health management.