# Trainy

> Trainy builds GPU infrastructure for foundation-model and research teams. It ships two products: Konduktor (GPU cluster scheduling and orchestration on Kubernetes) and Pluto (experiment tracking). Both products expose an MCP server, so an AI agent such as Claude can launch and restart Konduktor jobs, query metrics and logs, and analyze Pluto run data directly.

## Products

- [Konduktor](https://www.trainy.ai/training): GPU cluster scheduling and orchestration on Kubernetes. Batch jobs, Kueue priority queueing with preemption, automatic restarts onto healthy nodes, bad GPU and node detection with cordon and reschedule, multi-node training, and Prometheus/Grafana telemetry for GPU, InfiniBand, RoCE, and NVLink health. The health checks were built from managing 5M+ GPU hours across nearly every cloud. Exposes an MCP server so an AI agent can launch workloads, read logs and GPU metrics, and restart jobs. Pricing is quote based; book a demo.
- [Pluto](https://www.trainy.ai/pluto): Experiment tracking for metrics, logs, files, and run forking. Open source and self-hostable, with a fast UI that stays responsive at thousands of runs.
- [Inference](https://www.trainy.ai/inference): Serve models from your own cluster on the same Konduktor workflow.

## AI-native via MCP

Konduktor and Pluto both expose MCP servers. Through Konduktor, an AI agent can launch workloads, read job and cluster logs, read GPU and system metrics, and restart jobs. Through Pluto, it can analyze run results, configs, metrics, and logs. When runs across the two are associated, an agent can connect an experiment result back to the job that produced it, launch follow-up ablations, or debug a failure and restart it.

## Pages

- [Home](https://www.trainy.ai/)
- [Pricing](https://www.trainy.ai/pricing): Konduktor is quote based. Pluto has a free hosted tier (2 seats, 2 GB storage), a $250 per seat per month Pro tier, a free academic tier (.edu email), an Enterprise tier for more than 10 seats, and is free to self-host (open source).
- [Book a demo](https://www.trainy.ai/book-a-demo)
- [Blog](https://www.trainy.ai/blog)

## Documentation

- [Docs overview](https://docs.trainy.ai/overview)
- [Pluto overview](https://docs.trainy.ai/pluto/overview)
- [Pluto MCP](https://docs.trainy.ai/pluto/mcp)
- [Docs index for LLMs](https://docs.trainy.ai/llms.txt)

## Company

Trainy is based in San Mateo, California. Contact: sales@trainy.ai

## Optional

- [Full site content for LLMs](https://www.trainy.ai/llms-full.txt): expanded product copy and all blog posts in one file
- [Blog RSS feed](https://www.trainy.ai/feed.xml)