As a part of Kubernetes Book Club initiative we are currently exploring Machine Learning on Kubernetes. For reference we are using the book Machine Learning on Kubernetes but also planing to cover additional content to make sure we all understand all the basics and apply them in real-work scenarios. With our discussion in book club we created following list of topics to cover.
- Overview of Machine Learning
- MLOps Overview
- Components Machine Learning Platform
- Understanding and setting up Jupyter notebooks D
- Deploying JupyterHub on Kubernetes
- Getting familiar with Notebooks
- MLflow or Kubeflow ?
- Serving an existing model
- Understanding a business problem
- Data collection, processing and cleaning
- Performing exploratory data analysis
- Overview of feature engineering
- Building a Data Pipeline
- Building a Model
- Deploying/Serving the model
- Monitoring Models
- Configuring CI/CD models
Above may change based on the progress. As of now we have completed two sessions and following are recording and summary of both of them.
Session 1 : Overview of Machine Learning and MLOps Overview
In this session looked at how does AI, ML, Deep learning, Mathematics & Statistics are related. Then we discussed about how ML is just a of entire MLOps pipeline. We then covered different user personas like Domain experts, Data Scientists, IT Engineer, App Developers; who would part of the entire MLOps cycle.
Session 2: Deploying JupyterHub on Kubernetes and a sample Notebook Run
In this session looked at what Jupyter Notebooks and how to setup the JupyterHub on Kubernetes. We then briefly touch upon different Machine Learning methods like supervised, un-supervised and Reinforcement Learning and took an example of supervise learning from this repo.
In the coming sessions we’ll covering topics mentioned earlier in this post. All the of sessions are very beginner friendly, free and online. You an RSVP to the events here.
