Solving more problems is the key to improvement – Siddharth Chaurasia, Tata Consultancy Services. – INDIAai

Dr Siddharth Chaurasia is a Senior Scientist at Tata Consultancy Services.

He is an active researcher and frequent speaker on AI and how we can use it to improve the quality of life for humans.

INDIAai interviewed Siddharth to get his perspective on AI.

How did your journey with AI start?

Andrew Ng, through his Machine Learning course on Coursera, has taught millions, and I am one of them. In 2012-13 AI started to make a buzz, and I came across Prof Ng and fell in love with his lucid explanations and way of teaching the subject. As is often the case with kids who find their favourite subject and teacher he likes most, so was with me. Traditional degree programs followed it, but my start was with Prof Ng. 

You started as a software developer. You previously worked as a database administrator, software engineer, and technical project manager. How has your journey been?

When I look back at the path taken, it seems to be a well-carved one – from database developer and admin to data person and then how to analyze data to get insights and define actions. I have been around data since the start of my career. But this is something that happened rather than planned.

The final shift to AI and data science came through my master’s in data Analytics in 2014, and my doctorate followed.

What exactly do you do as a Senior Scientist (AI & Machine Learning) at Tata Consultancy Services? What are your everyday responsibilities?

My current role is at the crossroads of AI and Education. It is an endeavour for AI enablement of the enormous task force within my organization by designing and implementing AI programs for different job and role areas. With TCS, the challenge is scaling up rapidly to meet current technological demands. We focus on innovating AI-based products and solutions to make that happen and generating insights into Human Resources.

In addition, day-to-day responsibilities include: remaining abreast with the evolving organizational needs and marrying them with relevant technological trends. Going through the latest research and applying relevant SOTA to meet and improve the objectives. Another outcome of our work is to generate IPs and use them in our offerings.

What skills do you expect from a fresher interested in working in AI?

For freshers, basics come first – Algorithms, Data structures and a programming language. Then, a basic understanding of math, especially Linear Algebra, Calculus, and a few optimization methods along with Statistics, forms another set of basics. In essence, most of these are in their curriculum. 

If one is good with these, the next step is to get your hands in by working with the problems. It is an efficient field, and each situation tends to be different. Thus, spending time to solve as many as possible is the only Mantra to keep improving.

What kind of changes do you see in the world of AI over the next decade?

The pace at which AI is evolving makes this question fascinating. Frankly, even for an AI person, this question is tricky. Robots in Operation Theaters, along with Doctors to Passenger crafts piloted by AI where Pilot enjoys the flight, all are possibilities. Metaverse may come to life in a few years. If it lives up to expectations, it can take AI to a new dimension where both virtual and real co-exist. 

Like any technological innovation, AI has both bright and dark sides. For researchers like us who see AI as an aid to humanity, endless AI support can be in every field. 

Can you say something about your doctoral research findings?

My research is at the intersection of Finance and Machine learning. The research proposed and patented an ML-driven approach to minimizing risk and drawdowns in the portfolio. Another critical finding during this research was the proper presentation of data for the underlying objectives and their impacts. At the time of my research, the term data-centric AI was not common, but my work empirically established the importance of relevant data.

Is a degree in data science sufficient to land a job?

The best thing about the field of data science is anyone can contribute and can come up with cutting-edge work. Of course, a data science degree helps, but knowledge is the only necessary and good thing. 

Furthermore, the open-source nature of this field has provided opportunities for everyone to learn and work in the area. It is a field that has evolved rapidly. Hence the most important attribute will be the eagerness to learn and experiment. 

What advice would you give to someone interested in working in AI research? What are the best strategies for getting ahead?

AI is an applied field. So there are two ways one can proceed in AI research. Either work on one AI subfield like Computer Vision, NLP, or Timeseries and apply it to various domains. Or the other way can be to focus on the application domain and use the AI in it, e.g., in Healthcare, Finance, Education and so on. 

Core AI researchers often adopt the first approach. The objective is to pick an existing problem and work on its innovative resolution. If the impact of the work is visible, it keeps researchers motivated, primarily if it is in the form of value addition to society. 

When in research, one needs to be abreast with the latest work; hence keeping up to date with the research is a must. In addition, contributing to any open-source work which may align with one’s field helps in working with world-class researchers and learning and contributing to a cause. 

Could you please provide me with a list of important AI research articles and books?

I would recommend my favourites: 

1. Grokking Machine Learning by Luis Serrano

2. Deep Learning by Ian Goodfellow

3. Machine Learning Systems Design by Chip Huyen

For research articles, conferences like CVPR, ICLR, NIPS, ICML, and AAAI provide open and cutting-edge research. For breadth, we can again refer to the open-source repository arxiv, where we get early pre-prints.

Spread the love

Leave a Reply

Your email address will not be published.