Show HN: mlop – open-source ML Experiment Tracking

Hey HN!

We built mlop (https://mlop.ai), a fully open source (https://github.com/mlop-ai) ML experiment tracking platform, much like Weights & Biases.

Unlike the existing competitors we focus heavily on performance (yes our ingestion backend is in Rust), and top tier user experience, and we are fully open sourced with easy self hosting using docker.

Why did we build this? WandB was misleading about their performance, they say they are non blocking, but in fact, they block user code (see video https://docs.mlop.ai/docs/demo), our logger tries to be minimally invasive and performant.

WandB is a great product, and is far better than the competitors, we believe our product has a shot at tackling WandB (and almost certainly outdoes the competitors).

Migration from WandB is dead simple, our API is compatible with theirs, and we developed our own scripts to easily migrate all of your existing data to our platform (leveraging a dumb security issue that WandB has)

Would love to know your thoughts and feedback on this!