Describe your model lifecycle contributions: training pipeline design, experiment tracking, model versioning, deployment infrastructure, and production monitoring. Mentioning specific tools like MLflow for experiment tracking or Prometheus for model drift monitoring adds specificity that generic phrases cannot provide.

For research-adjacent roles at AI labs or large tech companies, publications may be expected in a separate section. For applied ML engineering roles at product companies, prioritize production deployment experience over academic output. A brief mention of a relevant paper is fine, but production impact should lead.

Include training data scale (millions of samples, terabytes of data), inference throughput (requests per second), and latency targets met. 'Deployed real-time recommendation model serving 50K predictions/second at under 20ms p99 latency' conveys production readiness in a way that model accuracy alone does not.