Frontier labs like Cursor, Meta, OpenAI, Anthropic, Mistral and Cohere all run continuous integration on Buildkite. On the inference side, teams building with vLLM, Coreweave and Anyscale push through the same system. Even heavy ML platform providers—NVIDIA, Lambda Labs, Hugging Face and Tecton—rely on it, as do applied AI services such as Harvey, Persona and Anyscale’s own apps. That breadth shows Buildkite handling everything from fresh research code to production workloads. The pitch is simple: spin up more than 100,000 concurrent runners, host through Buildkite or run agents in your own environment, and keep secrets and scale under your control. Buildkite handles orchestration, test scheduling and logs aggregation, while you decide where compute lives. If you want to see it in action, they’re at the AI Engineer World’s Fair in San Francisco from June 29 to July 2—and there’s an all-access trial you can sign up for today.