Choudhury Aniruddha - Continuous Machine Learning with Kubeflow

My beloved parents and family

Aniruddha Choudhury has 5 years of IT professional experience in providing Artificial Intelligence development solutions, MLOPS Kubeflow, Multi Cloud GCP, AWS, Azure and is passionate about providing Data Science and Data Engineering complex solutions in machine learning and deep learning. He is always looking for new opportunities for a new dimensional challenge for high impact business problems to become a valuable contributor for his future employers.

Presently he is working in Publicis Sapient as Senior Data scientist (Full stack MLOPS) for the last 2 year. Previously he worked with Incture technology and before that he has worked at Wells Fargo Bank on diverse financial products AI solutions on various lines of business.

As a tech geek, Aniruddha is always enthusiastic about working in cross-dimensional knowledge. He is working on Kaggle and Google data projects building a statistical and machine learning NLP model with an end-to-end data wrangling and preparation, building model framework with visualization and predictive/text/image analytics and Amazon AWS and Microsoft Azure DataBricks Cloud with Apache scala and spark and finding patterns in all forms of data. He has a passion to break complex problems in data science field and find resolutions with deep learning and machine learning. He is also working on deep learning frameworks like Tensorflow, Keras, and Pytorch. As an individual, Aniruddha always believes in constantly learning new skills and taking the road less travelled. He likes to keep himself updated about the technological world and is always toying with some new innovation.

He has mastered in building Artificial Intelligence solutions and finding complex patterns from research papers to gain optimal solution to current product development and possesses a self-innovative mind alongside.

Rajdeep Kumar is a lead data scientist at one of the top consulting companies. He has a Post Graduate Degree in Data Science from BITS Pilani, and is an alumni of IIIT Bangalore. He has more than 11 years of work experience in the IT industry. In his capacity, he drives and contributes to the Data Science activities along with defining the road map, scoping and mentoring the team members. He is an expert in deriving and productionizing end-to-end ML solutions to grow business and have a positive impact on the key business KPIs.

There are a few people I want to thank for the continued and ongoing support they have given me during the writing of this book. First and foremost, I would like to thank my parents for continuously encouraging me for writing the book I could have never completed this book without their support. I would like to thank my friend Dr. Someswar Deb, who is working as MSL in Novartis, for his constant support and motivation.

I am grateful to the journey provided from my companies who gave me support throughout the learning process of my career.

Thank you for all the hidden support provided. I gratefully acknowledge Mr. Abhishek Kumar, Senior Director of Public is Sapient for his kind technical help for deployment related stuffs which was helpful for this book. I would like to thank Sivaram Annadurai Senior Manager at Publicis Sapient for his constant motivation in technical and business aspects in machine learning real-time projects.

My gratitude also goes to the team at BPB Publications for being supportive enough to provide me quite a long time to finish the first part of the book and also allow me to publish the book in multiple parts. Since image processing being a vast and very active area of research, it was impossible to deep-dive into different classes of problems in a single book, especially by not making it too voluminous.

This book focuses on the DevOps and MLOps of deploying and productionising machine learning projects with Kubeflow in Google Cloud platform. The authors feel that in this era of machine learning, lot of companies failed to make production of AI/ML projects in real time which was also a study from Forbes. It is compelling and relevant content for todays practicing DevOps/MLOps teams as this sector is still changing. So, many machine learning platforms today take different approaches to the architecture and solution space of managing machine learning workflows. The core concepts of Kubernetes and Kubeflow and its architecture alongside teaches us how to approach and make your AI/ML projects from training to serving with scale in production with Kubeflow.

This book starts by taking you through todays machine learning infrastructure of Kubernetes and Kubeflow architecture. We then go on to outline the core principles of deploying various AI/ML use cases with TensorFlow training serving with Kubeflow and explain how Kubernetes solves some of the issues that arise. We further show how to use TFX with Kubeflow alongside Explainable AI for determining fairness and biasness with What-if Tool. We learn various serving techniques framework for different use cases with Kubeflow KF serving. After that we look at building sample computer vision based UI in streamlit and deploying that in Google cloud platform Kubernetes and Heroku deployment.

This book is divided into 8 chapters. They cover Kubernetes, Kubeflow basics, advance deployment projects with Kubeflow, AWS Sagemaker deployment and explainable AI with real time examples for deployment and container creation with Docker and building pipeline in Kubeflow. More interest will arise among learners in Machine learning deployment with Kubeflow.