Rust Developer
Resume Metrics

Lean descriptive. The gold standard is a hard number, though scope and direction stand on their own just fine. Name the first benchmark suite you stood up for your team, a crash-prone C service you turned memory-safe, or a hot path you drove from slow to instant. A recruiter credits each of those, and all of them are true. Every type here has a worked example further up.

Sure, as long as it holds true and you can back the claim. If a service ran roughly twice as quick after your rework but the benchmark went unsaved, a line like "cut the latency by half" is fair. Favor relative percentages where the real numbers stay under wraps. The one rule: you could talk an interviewer through the steps.

Resist it. A faked figure falls apart the moment someone probes it, and Rust numbers invite probing: an interviewer might push on the latency one or the safety one. Invent a stat and it can undo an otherwise solid interview. Give a qualitative point there instead; it rings true and still delivers the win.

Just the strong ones. Reserve a figure for a couple of your sharpest bullets from your most recent role, the ones a recruiter sees first. Slap a number on each and the sharp ones blur together, and filler creeps in. A handful of solid metrics beats a screen of weak ones.

Use whichever hits harder, kept honest. A large relative jump looks strong as a percentage ("cut latency 60%"); a large raw figure works fine on its own ("30k requests a second"). Skip a bare percentage carrying no baseline, since "improved speed 40%" invites the obvious question. Provide both where you have them: "cut p99 80%, from 200ms to 18ms."

Junior resumes want them too, and they crop up more often than you would guess. A latency number before and after, the number of features you shipped, the tests you added, or the panics you kept from production each come out of a lone project or internship. Revenue misses the point; evidence you moved something real is what counts.

Most is salvageable, honestly. An APM such as Datadog or Prometheus still holds your latency, error rate, and throughput; a profiler or flamegraph gives you the hot paths and allocations in moments; query times live in your slow-query log. Should the service be gone, give a rough estimate and say so.

Just one. A single figure up top, the latency you cut or your top win, earns another second or two of reading. Leave the rest to your experience bullets below. The Rust resume guide walks through that summary.