Skip to content

Guides

How-to guides for specific tasks in KITT. Each guide walks through a particular workflow -- from engine setup to production deployment -- with commands you can copy and adapt.

If you are new to KITT, start with the Getting Started section first.


Engine & Benchmark Workflows

  • Engines -- Set up, inspect, and configure inference engines.
  • Benchmarks -- Run test suites, create custom benchmarks, and interpret results.
  • Results & KARR -- Store, compare, import, and export benchmark results through KARR.
  • Campaigns -- Automate multi-model, multi-engine benchmark runs with campaign configs.

Deployment & Infrastructure

  • Docker Deployment -- Generate composable Docker stacks with kitt stack.
  • Web Dashboard -- Launch the web UI, configure TLS, and use the REST API.
  • Monitoring -- Deploy Prometheus + Grafana + InfluxDB monitoring stacks locally or remotely.

Distributed Execution

  • Agent Daemon -- Run a KITT agent on GPU servers for remote benchmark execution.
  • Remote Execution -- Set up remote hosts and run campaigns over SSH.

Automation & Integration

  • CI Integration -- Generate reports and post results to GitHub PRs from CI pipelines.
  • Plugins -- Install, list, and remove third-party KITT plugins.
  • Bot Integration -- Run a Slack or Discord bot that responds to benchmark commands.

Visualization & Hardware

  • Charts -- Generate quantization quality tradeoff curves and export data.
  • DGX Spark -- Notes for running KITT on NVIDIA DGX Spark hardware.