"Projit: An Open Source Python Tool for Decoupled Data Science"

on YouTube ↗11m

" summary

Projit is an open source python package that actually manages project metadata — datasets, experiments, executions, hyperparameters and results — with a local store that can live with code and be queried from a CLI. The presenter demonstrates working commands (status, list, register, finish), generation of execution IDs and recorded metrics like mean runtime, a CLI inspired by UNIX and git, templated project initiation, and concrete use in scientific papers enabling cross‑project meta‑analysis.

AI-written summary · about

" chapters (7)

1. 0:00Introduction
2. 0:47Decoupled data science motivation
3. 3:57Projit design
4. 6:26Metadata store and decoupling
5. 7:29Live Projit CLI demo
6. 9:39Projit as Python package
7. 10:23Conclusion

" related talks

• 13mPremiere - Conf42 Python 2023Conf42 Python 2023
• 26mCLI, CI, DIY: Crafting Python Based Dev Tools in a DevOps VacuumLiubov Burtseva, Uri Drori · PyCon Israel 2024
• 25mAutomated Scientific Pipelines With DoitSam Hames · PyCon Australia 2015 Science/Data Miniconf
• 54mHow to build stunning Data Science Web applications in PythonFlorian Jacta, Vincent Gosselin · PyCon US 2023
• 25mpypet: A Python Toolkit for Simulations and Numerical ExperimentsRobert Meyer · PyData Berlin 2016
• 30mEmpowered by Python - A success storyJens Nie, Peer Wagner · PyCon.DE 2017
• 45mOpen Source for Open Science with Frictionless DataLilly Winfree · All Things Open 2020
• 09mAccelerating ML Prototyping The Pythonic WayParul Gupta · PyBay 2024