Tech Screening

Inference batching

GPU utilization is low while request latency is high. How would you test whether batching or queueing is misconfigured?

Practice context

This tech screening exercise trains prompt interpretation, explicit assumptions, a concrete response, and a concise explanation of the decisions made. It is marked medium difficulty.

The signed-in workspace provides the tools required by the round, such as recording, code execution, diagrams, structured notes, or evidence panels. Attempts are saved so feedback and progress can be compared over time.

Evaluation uses task-specific criteria and the candidate seniority selected in the preparation path. The public page excludes reference solutions, hidden checks, evaluator instructions, recordings, transcripts, and private attempt data.

What this task practices

This tech screening exercise trains the same sequence expected in an interview: understand the prompt, make assumptions explicit, produce a concrete answer, and explain the decisions behind it. The task is marked medium difficulty. The private workspace adds the tools required by this round, such as recording, code execution, diagrams, evidence panels, or structured notes. It also preserves the attempt so later feedback can be compared with previous work. Evaluation is based on task-specific criteria and the candidate seniority selected in the preparation path. Public pages never expose the reference solution, hidden tests, evaluator instructions, or another candidate’s work.