Welcome to Hugging Science — part of the Hugging Face ecosystem dedicated to ML for science. Join our community building tools that enable more people to work on interesting scientific problems.
Follow these steps to submit a strong application. Start early — the more you engage with our community before applying, the better your chances.
Please include the following in your application:
Your proposal must include a well-defined schedule:
Become a member of an awesome community working on ML for science!
Make sure you meet the requirements and understand what's expected.
Keep these in mind throughout the application process.
These are suggested project ideas. You're welcome to propose your own idea — just discuss it with a mentor first!
It would be great to access MLIPs (Machine Learning Interatomic Potentials) directly from Hugging Face.
This project involves creating a common Python object to describe Systems, with two main possibilities being
torchsim.SimState or ase.Atoms.
Most protein generation pipelines are built with a sequence of diffuser, folder, and then scorer. We want those to be plug and play — making it easy to swap components and build custom pipelines for protein design workflows.
We want to be able to load FASTA, FASTQ, mmCIF, and HDF5 files quickly for machine learning. We've recently added some of this functionality, but there is so much more to be done and documentation is needed for all of it.