> For clean Markdown content of this page, append .md to this URL. For the complete documentation index, see https://docs.nvidia.com/dynamo/llms.txt. For full content including API reference and SDK examples, see https://docs.nvidia.com/dynamo/llms-full.txt.

# Performance Optimization Skills

These skills form the optimization workflow: capture a workload contract, benchmark a confirmed
baseline, then challenge it with one candidate at a time until the Service Level Objectives (SLOs)
are met or the budget runs out. The loop also uses
[`deploy-dynamo-recipe`](https://github.com/ai-dynamo/dynamo/tree/main/.agents/skills/deploy-dynamo-recipe),
listed on the [Deployment and Operations](/dynamo/dev/agent-skills/deployment) page, to deploy the confirmed baseline and
each approved candidate.

A prompt that reaches them: "Optimize this deployment for output tokens per second per user under a
200 ms time-to-first-token SLO. Budget 8 GPU-hours and stop after three failed deployments."

| Phase                   | Skill                                                                                                                           | Use for                                                                                                |
| ----------------------- | ------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------ |
| Workload contract       | [`synthesize-user-workload`](https://github.com/ai-dynamo/dynamo/tree/main/.agents/skills/synthesize-user-workload)             | Capture the workload requirements and immutable user-provided or confirmed baseline deployment.        |
| Baseline authoring      | [`author-baseline-dgd`](https://github.com/ai-dynamo/dynamo/tree/main/.agents/skills/author-baseline-dgd)                       | Draft a baseline when no catalog recipe matches, then present it for explicit user confirmation.       |
| Benchmark configuration | [`configure-aiperf-benchmark`](https://github.com/ai-dynamo/dynamo/tree/main/.agents/skills/configure-aiperf-benchmark)         | Freeze the AIPerf workload, objective, load policy, and Kubernetes execution manifest.                 |
| Benchmark execution     | [`run-aiperf-benchmark`](https://github.com/ai-dynamo/dynamo/tree/main/.agents/skills/run-aiperf-benchmark)                     | Launch, monitor, debug, and collect one run-scoped AIPerf benchmark.                                   |
| Results analysis        | [`analyze-aiperf-results`](https://github.com/ai-dynamo/dynamo/tree/main/.agents/skills/analyze-aiperf-results)                 | Validate and normalize results, evaluate SLOs, and compare compatible candidates.                      |
| Proposal selection      | [`consult-perf-knowledge`](https://github.com/ai-dynamo/dynamo/tree/main/.agents/skills/consult-perf-knowledge)                 | Select one evidence-backed optimization proposal from the performance rules and engine guidance.       |
| Candidate creation      | [`create-optimization-hypothesis`](https://github.com/ai-dynamo/dynamo/tree/main/.agents/skills/create-optimization-hypothesis) | Apply only the selected change to create a challenger-ready deployment draft.                          |
| Adversarial review      | [`perform-adversarial-review`](https://github.com/ai-dynamo/dynamo/tree/main/.agents/skills/perform-adversarial-review)         | Review a candidate for correctness, comparability, feasibility, attribution, and worthwhile GPU spend. |

See the
[optimization loop](https://github.com/ai-dynamo/dynamo/blob/main/agent-docs/guides/optimization/optimize-loop.md)
for the full sequence and the
[evidence rules](https://github.com/ai-dynamo/dynamo/tree/main/agent-docs/rules) for benchmark
validity requirements.