Abstract: 

There are very few open-source tutorials/resources that cover the end-to-end of what is required to build an enterprise level deep learning pipeline for Computer Vision related solutions.

There are multiple components that go into building an end-to-end solution for Computer Vision. All these are already available as open-source projects but are disparate and require an expert to leverage them well.

This tutorial aims to bring together all such components and make them work together as we build and end-to-end pipeline which the audience can use for their organization's Computer Vision related projects.

The components that we will discuss:
* Interactive Ground Truth Data Annotation
* Model training
* Building lighter models for fast inference
* Metrics visualization
* Pilot phase
* Deployment

The tutorial will cover the above aspects as we demonstrate live on a particular project to the participants.

All participants will have access to the code/notebooks that will be discussed during the presentation so that they can leverage them offline.

Session Outline
* Interactive Ground Truth Data Annotation
* Model training
* Building lighter models for fast inference
* Metrics visualization
* Pilot phase
* Deployment

Background Knowledge
Python and familiarity with Machine Learning Life Cycle

Bio: 

Nilav is a Manager, Data Scientist in Optum with a focus on architecting and deploying ML models at enterprise scale – on premise and cloud. He has over a decade of experience in designing and developing engineering and AI solutions in finance and healthcare industries.

In his stint at Optum, he has worked on productizing deep learning models in the areas of computer vision and NLP and has filed 4 patents in these areas. He holds an M.S in Computer Science from Georgia Tech. His other passions are consulting on system & architecture design and technology training. He has trained 1k+ engineers around the globe on Python and Machine Learning. In his free time, he loves developing his chess skills.