Design a Streaming Job Scheduler

Design a task scheduling service that continuously ingests tasks and dispatches them to workers. Each task has: - `id`: unique task identifier - `deadline`: execution deadline - `prerequisites`: a list of task IDs that must complete first The system should: 1. Accept tasks in batches or as a stream. 2. Validate task definitions, including malformed input, duplicate IDs, and invalid prerequisite references. 3. Verify that task dependencies form a directed acyclic graph. 4. Dispatch only runnable tasks, where all prerequisites are completed. 5. Among runnable tasks, prefer the one with the earliest deadline. 6. Support multiple workers consuming tasks concurrently without assigning the same task twice. 7. Handle worker crashes, retries, and recovery. 8. Scale to large task volumes and provide useful monitoring. Discuss the API design, storage model, scheduling logic, DAG validation strategy, concurrency control, fault tolerance, and scaling approach.