Hey HN! We're Roanak and Andrew from Trainy (YC S23). We build GPU infrastructure for ML teams (scheduling multi-node training jobs on Kubernetes). When Neptune announced they're shutting down, our customers didn't have a clear path forward. The alternatives exist but none of them matched the UI experience Neptune had, especially at scale. So we decided to build Pluto.
Pluto is an open-source experiment tracker based on our fork of MLOp. The main idea is that you can add one import alongside your existing Neptune code and it logs to both platforms simultaneously. You validate that everything matches on real training runs, then when you're ready, set an env var and all Neptune API calls redirect to Pluto. We also built a Neptune exporter for historical runs.
We're focusing heavily on having a UI that stays responsive at scale, since ML teams can have thousands of runs per project and the tracker is open all day. If you find anything slow or buggy in the playground, we'd love to hear about it.
Beyond the compatibility layer, we're working on: tensor logging with on-the-fly visualization (log raw tensors instead of pre-rendered plots), code diffing between runs, and Linear/Jira integration. We just shipped Pluto MCP in alpha, which lets you query experiment data with an LLM.
Pluto is an open-source experiment tracker based on our fork of MLOp. The main idea is that you can add one import alongside your existing Neptune code and it logs to both platforms simultaneously. You validate that everything matches on real training runs, then when you're ready, set an env var and all Neptune API calls redirect to Pluto. We also built a Neptune exporter for historical runs.
We're focusing heavily on having a UI that stays responsive at scale, since ML teams can have thousands of runs per project and the tracker is open all day. If you find anything slow or buggy in the playground, we'd love to hear about it.
Beyond the compatibility layer, we're working on: tensor logging with on-the-fly visualization (log raw tensors instead of pre-rendered plots), code diffing between runs, and Linear/Jira integration. We just shipped Pluto MCP in alpha, which lets you query experiment data with an LLM.
Live playground (no signup): https://demo.pluto.trainy.ai/o/dev-org/projects/my-ml-projec...
Quickstart: https://docs.trainy.ai/pluto/quickstart
Listed on Neptune's official transition hub: https://docs.neptune.ai/transition_hub/migration/to_pluto
Let us know what you think! We'd especially love feedback from anyone managing experiment tracking across a team.