Sunil Chathaveetil was part of the founding team at Mintmesh and is currently its CEO. He builds and nurtures the organization into a client-centric, value-driven software provider.

Sunil did his electrical engineering at the National Institute of Technology in Jaipur, India. He has earned a Master's in Software Engineering from Central Michigan University and Executive Leadership stripe from Cornell.

INDIAai interviewed Sunil to get his perspective on AI.

As an electrical engineer, what inspired you towards AI?

​​As an electrical engineer from one of India's premier engineering colleges, NIT Jaipur, I landed my first job at Crompton Greaves, a large OEM for Electrical equipment and related services. Even at that time, around the early 2000s, it was becoming apparent that technological innovations rapidly influenced energy creation, transmission, distribution, and consumption. I was extremely interested in the innovation that was appearing over the horizon in this industry and the tentative baby steps being taken by industry veterans to embrace it. 

Over the years, I became more interested in software engineering, and my life as a technologist began to take shape. I saw the huge disruption happening in various sectors like banking, finance, manufacturing, and operations with new tools and ideas that changed how work was processed. In all this, the one area which appeared to be the least served by these more recent technologies was the engineering industry. Engineers could be better served if they were liberated from mundane, tedious tasks and given opportunities for more impactful engineering activities. That's my inspiration to use AI to better the lives of my fellow engineers.

Having worked in this field for over a decade, why must AI platforms and services be Responsible and Ethically Aligned in today's world?

Any solution wins or loses on the merits of its adoption. Therefore, AI solutions need to be considered responsible and ethically sound for worry-free adoption by the user community.

In today's world, communities will use AI platforms and services that they believe deliver business and societal value and include trust, transparency, and fairness in their results. In addition, they look for easy explainability, rigorous accountability, industry-grade security, privacy, and regulatory compliance in these AI tools.

Responsible and Ethical AI helps drive fairness, despite biases baked into the data, and gain trust through transparent, explainable AI methods. These AI tools and services often replace or augment human decision-making and amplify good and bad outcomes. Responsible and explainable AI enables proper outcomes by resolving risk versus value dilemmas.

What prompted the formation of the Mintmesh Corporation?

Catalyzing technological change in the engineering community became a matter of faith for me. So I decided to walk the talk and founded a company with other like-minded engineers to create a solution where none existed. So here we are with our company Mintmesh and its flagship tool, Rudy for Engineers, the world's first and only technical evaluation platform powered by explainable AI and built entirely by our AI lab in India.

How are you incorporating data analytics, AI, and ML into your organization?

Data analytics, or actionable insights, as we define it, is a key element of our solution, and the instrumentation for it is a core component of our technology stack. AIML techniques are built into our software offering to create an entirely new genre of Language learning called Engineering Language processing (ELP). As you know, AI modelling and training require considerable data and context analysis. Our organization has built training algorithms and ELP models to securely consume, process, and contextualize incoming data streams. Our models are very data-centric, driven more by enriching and enhancing data. It is combined with Regular micro-tuning using function-specific microservices for contextual training of the model and buildup of the ELP libraries, ensuring increased accuracy of the AI model. 

What obstacles have you encountered, and how have you surmounted them?

Pioneering AI in a field as complex as procuring Engineering equipment and technical assets had its challenges. Our primary obstacle was that no libraries existed in the natural language processing world to understand and process engineering language and technical specifications. There also needed to be standardized knowledge of engineering assets to solve this complex problem of reading and understanding technical documents. It was an initial setback for us, and we decided to create our proprietary engineering language processing capabilities. 

We took nearly 18 months to build a viable model that worked across different conditions. After having built that, the following main challenge was the availability of datasets that could be used to train these models and the ability to derive unbiased data of high quality. After building a complex ELP capability, we built training algorithms that could work on small data to achieve a given accuracy goal. As we matured with the trained data model, we now have a comprehensive solution for most technical equipment encountered for large Cap projects across the engineering Procurement and Construction Industry.

Please give us a glimpse of the data science ethos at Mintmesh Corporation.

As our AI models are considerably data-centric and play a significant role in selecting and procuring Technical Equipment, our AI models must be explainable and address the needs and interests of the human in the loop. Our design principle is centred around Human in the Loop AI solutions (HITL). We use traceability techniques like explainable AI, which helps build trust between the system and the human. The data science ethos within Mintmesh's innovation centre strongly reflects the need for the ethical and unbiased application of data science in data discovery, data sampling, acquisition, data screening and data security.

How do AI and data science fit into your company model?

The nucleus of our solution is built around a data-centric AI model using ELP and related techniques. Our fundamental model is derived from the concept of explainable AI, which adds additional responsibility to our data science team to ethically curate and process data, including safe acquisition, data qualification and bias reduction and secure handling and disposal of such data.

What advice would you offer to someone interested in pursuing a career in AI research? What should they focus on to progress?

AI, as a subject of research and application, has gone beyond the realms of the lab and is now widely used to solve real-world problems. We can see a rapid coalescing of career opportunities around three domains – Data Management, Model Engineering and AI DevOps. All three are essential pillars of the AI world. For folks exploring ways to enter this world, they can begin with any of these pillars to make their initial foray. To answer the second part of this question, I would say that the most exciting new phase of AI work will be the evolution of Transformers (like ChatGPT, BardAI etc.) Research scientists can discover and invent new methods to solve complex problems in their expert domains, perhaps something in gene therapy or new energy.

What books and other intellectual works have had the most critical effect on your life?

It is a great question that lends itself to multiple conversations due to the various roles I have chosen to play during my career. However, through all these journeys, the few constants that have anchored me are the "Bhagavad Gita" and the book on "Self-Unfoldment" by Swami Chinmayananda. The teachings have helped me manage change, endure life-altering events and pursue excellence, free of expectations.