"It’s all about bringing in domain knowledge" 

Disruptive improvements in molecular imaging. Diagnostics for power generation. A planning system for an optimal power flow in grids that deal with a rising share of renewables. Or an optimized energy management for buildings. All these innovations have been made possible using Artificial Intelligence (AI).  Amit Chakraborty is an AI innovator and strategist that has contributed to solutions in all these fields. He has dealt with data throughout his entire career. For his Lifetime Achievement he is awarded the Siemens "Inventor of the Year 2021". 

"I have been working with AI for over 20 years now", Amit says, and his inventions have contributed to solutions in a variety of different fields. They span the entire Siemens universe, including improvements in molecular imaging or optimizing the energy management of buildings. He works as a Principal at Siemens Technology in Princeton, United States. "In my team, we make overarching use of AI. The common denominator is that – no matter from which source the data comes – we combine it with domain knowledge from Siemens to create smart and novel solutions. We combine the physical and digital worlds, real and simulated data." 

Amit refers to his specialization as Physics-aware AI. "Physics-aware AI at a high-level means using domain knowledge as a way to improve AI-based solutions and services. Often, the underlying physics can be described in the form of differential equations – within a data-driven modeling framework." According to Amit, that reduces the data requirement and improves generalization. "Encoding domain knowledge in our AI solutions gives Siemens a competitive advantage over our competitors, in particular the hyper scalers." 

When Amit is reflecting on his career, he always emphasizes the importance of teamwork. Or, as he formulates it: "I have a passion for developing people. And I could not have achieved any of my patents alone." The people he works with are as diverse as the fields of innovation that he has contributed towards. They come from different backgrounds, each of them has their specialization, some of them just graduated, and others have accompanied Amit on his journey for many years. 

There is a short moment in a joint team video call that tells a lot about Amit. Sridharan Palanivelu is being asked for how long he has been working with Amit, and – smiling – he says: "For more than 15 years! When I started this job, I also was new to the country. On the first weekend, Amit took me around and helped me settle down. I will never forget his help." Amit describes working for Siemens Technology as "very enriching": "You get to work with so many groups – it's very exciting to learn from such different perspectives." 

But the teamwork inside Siemens is not all. Amit is continuously building alliances with partners in diverse markets and publishing papers in scientific journals. He is an active member of the academic community – and his interest in scientific publications is a deep-hearted one: "I really like to read papers. It is one of my favorite activities, besides playing chess or following the news in politics and finance." He sees his professional evolution going along with the evolution of Siemens and its customers: "I have worked across many groups, and my work has always been inspired by our customer needs."

His recent patents and patent applications cover the emerging fields of Deep Learning and Hybrid AI, including applications in Control and Robotics, Simulation and Additive Manufacturing. One of his current topics is a project for Siemens Digital Industries Software. "We try to understand how energy can be optimally used for cars with electric drives", Amit explains, and one can see the sparks of curiosity in his eyes while he is talking. "Given that electric cars have a limited amount of battery charge, we are asking ourselves: How do you ensure passenger comfort and still maximize the range? Or said simply: How could we make it possible, that you turn on the heating in winter without significantly reducing the driving range of your car?" Yet another question of combining AI with domain knowledge.