Design Work Orchestration for Machine-Learning Data Pipelines
Company: Mercor
Role: Machine Learning Engineer
Category: ML System Design
Difficulty: medium
Interview Round: Onsite
Overview: Design orchestration for versioned machine-learning data workflows built as task DAGs. Cover durable scheduling, retries, worker loss, lineage, reproducibility, validation gates, atomic dataset publication, backfills, resource-aware compute pools, tenant fairness, and prevention of invalid training data.