#mlops-zoomcamp

Recently I have started learning about MLOPs in #mlops-zoomcamp course at @datatalksclub. MLOPs refers to the set of tools and best practices for development and operation of machine learning models. Till now, I have been learning and practicing how to develop machine learning models, but those models are not developed for just learning and research purposes. Those models must be put into production so that the end users can leverage it and use as per their needs and cases. MLOPs help with the production and operationalization of machine learning models similar to DevOps helping with software operationalization.

The course consists of 7 modules starting with model development in module 1 and covers almost everything from ml pipelines, model deployment to monitoring followed by an end to end project inorder to receive the certificate of completion of the course. I am very excited to learn and expand my knowledge on how we can actually create value by leveraging machine learning models for real world applications.

The first week is focused on setting up the environment and model development part where we develop models that we will be using throughout the course for operationalizing and putting into production. Some of the new things that I learned during the first week are described below:-

1. parquet file format contains less disk space than csv file for the same amount of records it contains.

2. There is another way of performing One Hot Encoding besides OneHotEncoder, using DictVectorizer.

So this is all for week 1 and I'm looking forward to learning new things and feeling really excited. The course is a real gold mine for getting hands on into MLOPs as the instructors are very well versed and proficient on the topic. The link to the course is https://github.com/DataTalksClub/mlops-zoomcamp. Happy learning!