Freelunch is an early-stage startup making declarative MLOps tools. Our goal is to make make ML work as expected in production.
We currently make 2 products:
-
Freelunch Open Source: open source framework for fabrication and improvement of enterprise-grade internal ML Platforms.
Instead of reinventing the wheel, it is designed to leverage existing best-in-class (open-source and proprietary) tools; without being tied to any specific tool (tool-agnostic).
Note: not an ML Platform (like MLFlow or Kubeflow); allows you to make your own internal ML Platform.
-
Freelunch Copilot: SaaS ML Engineering co-pilot that works on top of Freelunch Open Source.
It works with the ML team just like it was another human ML Engineer (e.g., makes PRs, interacts with team members via slack, builds datasets & models, operates the system, etc).
You can checkout our docs in this repository to understand better our vision for the evolution of MLOps.
- Checkout our Design Partnership Proposal Slides for a concise and visual explanation. This slide deck was originally built for bringing potential customers as Design Partners.
- Checkout our Design Notes for an in-depth explanation of not only Freelunch, but also of ML in production (including LLMs).
The slides are easier to understand, more polished and up-to-date, so I'd recommend them first. The Notes have way more stuff, useful if you want more details.