Skip to content

Migrate to Ray Serve and capability-based inference API - #45

Merged
jrabary merged 14 commits into
mainfrom
ray-serve
Mar 23, 2026
Merged

Migrate to Ray Serve and capability-based inference API#45
jrabary merged 14 commits into
mainfrom
ray-serve

Conversation

@jrabary

@jrabary jrabary commented Mar 19, 2026

Copy link
Copy Markdown
Member

Summary

  • replace the legacy FastAPI/Celery/provider stack with the Ray Serve + Typer runtime
  • make model capability derive from the base model class instead of task strings in config and metadata
  • simplify inference routes to /inference/segmentation/, /inference/detection/, /inference/tracking/, and /inference/vlm/
  • align the client, docs, examples, and tests with the new Python-config-only contract

Breaking changes

  • remove ModelConfig.task and the legacy task utility layer
  • replace model metadata field task with capability
  • remove /tasks/... endpoints and the client instance_segmentation() helper
  • require HTTP-exposed custom models to inherit one of the supported capability base classes

Validation

  • uv run pixano-inference --help
  • uv run pytest -q
  • uv run mkdocs build --strict

@jrabary
jrabary merged commit 1ab6f87 into main Mar 23, 2026
9 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant