Vidster.ai Release Notes

Vidster.ai SaaS now has a production-oriented foundation: authenticated users, Postgres-backed job records, local upload storage, and artifact downloads through protected API routes.

See docs/videolingo-saas-development-plan.md for architecture and deployment notes.

The worker layer now wraps the original Vidster.ai pipeline with a durable SaaS control plane. Each worker claims jobs atomically, sends heartbeat updates, reports events, and writes terminal state through authenticated worker APIs.

Local testing supports one worker per A5000 GPU; production workers can be deployed independently on RunPod or another GPU runtime.