Design a Work Orchestration System | Dataford Interview Questions

Your question is Design a Work Orchestration System. Take a moment with it on the right.

Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).

Design a work orchestration system.

Asked in the Onsite system design (45 min) stage. Frontier Data MLE track; named in thread replies ("Work Orchestration (Frontier Data)").

Address the following in a practical design:

Define the work, dependency, scheduling, priority, and state models.
Design APIs and the control plane for submitting, assigning, retrying, canceling, and inspecting work.
Explain worker coordination, durable state, leases, idempotency, backpressure, and failure recovery.
Show how the system handles ML workloads with heterogeneous CPU, GPU, memory, and data requirements.
Define observability, capacity planning, and operational safeguards.
Discuss how you would test the design and evolve it from a small deployment to a large distributed system.

Your question is Design a Work Orchestration System. Take a moment with it on the right.

Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).

Design a work orchestration system.

Asked in the Onsite system design (45 min) stage. Frontier Data MLE track; named in thread replies ("Work Orchestration (Frontier Data)").

Address the following in a practical design:

Define the work, dependency, scheduling, priority, and state models.
Design APIs and the control plane for submitting, assigning, retrying, canceling, and inspecting work.
Explain worker coordination, durable state, leases, idempotency, backpressure, and failure recovery.
Show how the system handles ML workloads with heterogeneous CPU, GPU, memory, and data requirements.
Define observability, capacity planning, and operational safeguards.
Discuss how you would test the design and evolve it from a small deployment to a large distributed system.