mlop

mlop is an open-source, performance-focused experiment-tracking platform for training machine-learning models, positioned as a Weights & Biases alternative with full API compatibility for easy migration. It offers real-time logging, experiment versioning, Git integration and multimedia tracking, and can be self-hosted via a Docker-based open-source architecture. Founded in 2025 by Stephen Sun and based in London, the company is backed by Y Combinator. It targets ML engineers and research teams who want a faster, self-hostable tracking tool compatible with their existing W&B code.