Sukanya Randhawa is a research scientist at Auroville consulting. She creates cutting-edge digital planning tools that combine satellite photos and public databases with GIS and machine-learning algorithms to produce insightful data.

She is currently working on a digital tool called LifeLands (LiLa) with Auroville Consulting (India) to develop nature-based solutions for climate action and sustainable development.

INDIAai interviewed Sukanya to get her perspective on AI.

How did you begin your AI journey at IBM? What was your role? Could you tell us about your responsibilities and roles?

At IBM, I worked as an AI research scientist focusing on research and development in areas of precision agriculture using Machine Learning tools. Led the expansion of crop classification products, part of an AI-based platform for agricultural analytics launched by IBM to drive IBM's agri-business analytics sector. With advanced AI tools and optical physics considerations, solutions targeted precision agriculture using geospatial data curation. Solutions varied from crop identification and acreage estimation to insurance validation and credit scoring. 

Apart from that, I led experiments and technology innovation toward developing novel IoT-controlled optics to boost solar yield for commodity photovoltaics at a reduced Levelized cost of energy to promote the renewable sector (solar) in India and globally.

During this time, I also created and led research and development of AI-informed geospatial technology ("Bluewater EYE") for monitoring and managing water pollution over vast bodies of water and over long timescales. This work was done in collaboration with IBM Watson Lab and the University of Chicago (Pf. Supratik Guha's Water-to-Cloud initiative) and involved mapping and study of the major rivers of India (Ganges, Yamuna, Godavari). My work included showing how we can accurately map multispectral satellite image data on river turbidity, CDOM, Chlorophyll, DO, EC, pH, Nitrate, Temperature, and Tryptophan. The work has been presented at several global platforms.

It's great to hear that you're using satellite images, public datasets, and AI to generate data-driven insights that enable intelligent decision-making for solar energy, sustainable water management, and ecological restoration. Can you tell us about your current projects?

Given my passion for Sustainability and Climate Action, I have always felt compelled to work and develop innovations for the environment and social good. I started working with Auroville Consulting about two years back, given Auroville's international acclaim for its social and environmental sustainability efforts.

I have developed a frontier digital planning tool called LifeLands (LiLa) that uses earth intelligence for responsible land-use planning, ecosystem restoration, and climate action. The platform includes four different products for solar energy, sustainable water management, ecological restoration and integrated land-use planning. LiLa creates critical data-based insights and visualization to support decision-making via multilayered information, including earth observation data, machine learning and in-depth subject expertise. It is a very comprehensive tool that has the potential to address multiple social and environmental problems at the same time. These include climate, energy and water security, sustainable rural development, poverty alleviation via livelihood generation, biodiversity conservation & natural ecosystem preservation. 

A tool like this could be a game-changer for India by providing rapid climate intelligence for land resource management and sustainable development and creating a solid impact at scale.

You have substantial research experience in a variety of countries. In comparison to international research, what do you believe needs to be improved in Indian research?

I feel concerned that given the overall potential in the country, the number of reputed institutions is far too low. Also, we should give science research higher importance than applied engineering as that fosters innovation and is far more prestigious globally. It's surprising that more than half of the engineers this country produces end up selling soaps (metaphorically speaking) and consider science and engineering only as a stepping stone to other better-paying professions. With a change in education system design, we could avoid this kind of national waste (for institutions and individuals). 

And I think we should not only focus on science and technology research, but also invest national funds for research in arts, sports and other social areas. We need to bring more diversity into our research mindsets. Besides, arts and science have been part of the "ancient" holistic Indian scientific thinking. To foster more balanced innovation and knowledge, we should bring back that confidence in these areas through research.

Also, focus on top-down approaches rather than bottom-up approaches should be adopted to teach and train more independent minds.

What advice do you have for those who aspire to conduct research in artificial intelligence? What must they concentrate on to continue?

I think one could contribute to the area of AI only if one has a solid foundation in mathematics and programming. It is such an evolving field that one should read, ingest and understand new kinds of models and how we could use them given a specific context. Also, I feel it is always beneficial if one comes from a different field than computer science as you bring a fresh scientific perspective for building solutions. AI is a tool that finds applications in numerous areas today. It is always good to develop expertise in overlapping/sister fields of Computer Vision, Robotics, Geospatial Analytics, IoT, etc. They pair excellently with AI skills. One piece of advice would always be to keep your mind open to new ideas, challenges and innovations that would help you to enhance your research quality and success.

What hurdles did you experience as a geospatial/AI scientist? How did you overcome the challenge? Can you share your insights with us?

Many years back, AI was a new field for me, but I have always considered ML/DL as new evolving tools— and you can get very comfortable with them given a good background in Mathematics and Computer science. And physicists/scientists usually have an acquired knack for gaining a good understanding of any subject in depth. 

I have always believed in learning by doing - I mainly learned through my projects, and it always helped me to read and interact with my colleagues who were AI experts. Also, thanks to the fantastic amount of online material - be it AI courses, Medium Articles, or Git-hub code sharing …I always felt that getting started was never a problem. Most of the issues were just a google search away. One learns other finer details by experience, and the complexity of the problem one tries to solve.

What is the one aspect that a researcher must never sacrifice?

There is never just one aspect. Every researcher knows that a PhD is nothing short of a metamorphosis towards the end. And no transformation can happen without due dedication, the ability to take constant failures in your stride and having that quality of never giving up on difficulties. You need to be mentally tough and very patient to do well in the long run in research!

Can you list the online resources that motivated you as a researcher?

I strongly support keeping yourself updated with the latest research through scientific publications and the more informal yet informative articles/blogs on various online platforms such as Medium, Kaggle, GitHub etc. That most of the new and evolving technologies are open-source– is a beautiful and powerful boon in today's age –and opens the doors of knowledge for every interested individual.