← All guides

How to hire inference engineers

Suppose your team serves an LLM and long prompts are delaying shorter requests. You want someone who can change the serving engine and explain what happened to latency under load.

What to put in the brief

Ask for a shipped change to scheduling, KV cache management, or serving metrics. Treat vLLM and SGLang as examples of relevant systems. Ask separately about production on-call ownership.

Three public contributions to inspect

We selected these merged vLLM pull requests by hand. Each credits its author. Availability and interest in a role are unknown.

What a hiring manager still needs to ask

These patches establish authorship of specific changes. For a role that owns a live fleet, ask for a rollout the person handled, the latency or error signal they watched, and the threshold that would have triggered a rollback.