Hypergol is a Data Science productivity toolkit to accelerate small team projects into production at the shortest possible time while still maintaining high standard of code. This is achieved by autogenerating structure and code in a form that is easy to extend, so only the project-specific code needs to be written.

Hypergol provides parallel execution capabilities with little restrictions. This enables a small team to accelerate time-consuming tasks by renting larger instances from a cloud provider, but with no additional infrastructure, reducing the time and cost of setting up.

The toolkit provides a standard serialisation (autogenerated) and storage system that natively enables parallel processing. The resulting data is also easily accessible from any python code, including jupyter notebooks.

Every Hypergol Project is connected to a git repo that acts as a version controller. Both code and generated data are linked to git branches and commits. Full data lineage can be retrieved, and the storage system is verifiable through SHA1 hashes.

Tensorflow model stubs can be generated that enables standardised model development and training that both connect to the storage system above, observe SOLID principles and enable autogenerated deployment with FastAPI.

See documentation for further details at: https://hypergol.readthedocs.io/en/latest/

Join our community at our Discord server.

Install it with:

pip install hypergol

Create your first projects:

python3 -m hypergol.cli.create_project MyFirstProject
cd my_first_project

And follow instructions in the projects README.md or the tutorial at: https://hypergol.readthedocs.io/en/latest/tutorial.html

The authors wish that Hypergol frees you from tedious tasks so you can focus more on the core part of Machine Learning and generate everything else.

Join our community at our Discord server.

All feedback is welcome at: [email protected]