Description:

Training a machine learning / deep learning model is one thing and deploying it to a production is completely different beast. Not only you have to deploy it to a production, but you will have to retrain the model every now and then and redeploy the updates. With many machine learning / deep learning projects / POCs running in parallel with multiple environments such as dev, test prod, managing model life cycle from training to deployment can quickly become overwhelming. In this talk, I will discuss an approach to handle this complexity using Docker and Python. Rough outline of the talk is,

• Introduction to the topic
• Problem statement
• Quick introduction to Docker
• Discussing the proposed architecture
• Alternative architecture using AWS infrastructure
• Demo

Prerequisites:

• Basic Python
• Basic Docker

Content URLs:

I will share the slides on my github repo for the evaluation by the team in some days. Other content will be shared on github after the talk.

Speaker Info:

My name is Saurabh Deshpande. I am working as a Senior Software engineer at SAS Research and Development centre, Pune. I have been using python since last three years and also teaching python in my company. I have more then 11 years of experience in architecture and development of enterprise scale web applications, cloud technologies such as AWS, OpenStack, CloudFoundry, server less and microservice based architectures. Since past three years I have been exploring and experimenting in the field of visual analytics, machine learning, deep learning using python based libraries such as pandas, scikit learn, pytorch and tensorflow.

Speaker Links:

https://www.linkedin.com/in/saurabh1deshpande/