Automate Machine Learning tasks with Github Actions





Automate your ML Experiments

In this presentation we will talk about an interesting marriage of Open Source pipelines and Machine Learning. In this age of automation, Github Actions does create a lasting impact for Deep Technologies . Thus was born MLOps. In this presentation we will be demonstrated on how to trigger Machine Learning workflows using commenting on an issue or pull request. We will also see how these workflow notifications can be mirrored to Discord through bots. This proposal aims to provide the audience some handy skills in automating their Machine Learning experiments.

Ever wanted to keep changing the regularization parameter and see how your model fares? Well now you can. Infact this short talk will come in handy if you want to know how to fetch performance metrics, diagrams and experimental plots.



Basic hands-on experience with scikit-learn based machine learning, Basics of Tensorflow/Pytorch or any other DL framework, Basics of Docker

Plus: Experience writing config yml files, cron services and Github Actions

Plus Plus Experience with ML automation (Weights and Biases , but not just that)

Video URL:

Content URLs:

Slides will be updated here : Link to Slides

An example proof of work (Forkbabu is my another account): My talk at SillyHacks 2020

Speaker Info:

I am Sayantan Das ,currently working as a researcher under CVPR Unit,ISI Kolkata. I pursued my bachelors in Computer Science & Engineering from West Bengal University of Technology and graduated in July 2020. I am a mentor for OpenMined. I previously mentored for Google Code-In 2019 in Tensorflow.Previous summer I did a research internship at Space Applications Centre,ISRO Ahmedabad under Microwave Data Processing Division Lab working on Machine Learning tasks in the Microwave Remote Sensing domain. Reading,Reviewing and Reproducing interesting research papers is a major hobby.

Speaker Links:



Previous Talks

For a sneak-peek at the slides or anything you want to know more , DMs are open at My Twitter Handle.

Section: Developer tools and automation
Type: Talks
Target Audience: Intermediate
Last Updated: