Meet Ashutosh Saxena: The serial entrepreneur proving robots can learn

Ashutosh Saxena M.S. ’06 Ph.D. ’09 built his career proving that robots can learn. From conducting early artificial intelligence (AI) research at Stanford to founding companies, Saxena now builds foundational models for physics with his company, TorqueAGI.

“My career has been observing people fight against the impossible. The real insight is when you believe that is possible, reasoning from first principles,” said Saxena. 

Saxena is a serial entrepreneur. He co-founded Katapult, a publicly listed fintech company; Brain of Things, a smart home technology company; Caspar AI, a healthcare company focused on early incident and disease detection and most recently, TorqueAGI.

When Saxena started programming as a teenager, he felt it was a brittle way of manually encoding things, he said.

“I wanted to see how we could get away from writing specific computer programs for robots, and instead make them learn by example,” said Saxena. 

However, the problems he wanted to solve, such as self-driving and unloading objects, could not be easily solved by writing manual computer programs.

Saxena turned to AI as the solution. 

“Working with my Ph.D. advising team [Andrew Ng, Sebastian Thrun and Daphne Koller] changed my perspective,” said Saxena. “I started looking at programming being data-driven. This might sound obvious now, but wasn’t the case in 2004.” 

The people working with Saxena recognized his talent to distill problems down to the basics and build up from there. 

“Dr. Saxena is incredible at tackling real-world problems with strong, technical solutions,” said Aditya Jami, long-time collaborator of Saxena. “Building models so that robots can do the work that humans cannot, such as unsafe work conditions, is a noble pursuit.” 

Throughout his journey, Saxena has also discovered an alignment between research and entrepreneurship.

“[Research and entrepreneurship] are very similar,” said Saxena. “You are picking up an impossible problem, and most will not think it is a good idea.” 

Saxena seemed to have a knack for tackling these impossible problems.

“We proposed that a robot can find out the distance to an object from a single image. Despite doubts, we did it,” said Saxena. 

“Dr. Saxena has the right background to deliver physical AI solutions into a variety of markets,” said Corey Gates, the CTO of Caspar AI. “He brings the right level of both technical knowledge and customer-facing expertise to create products that meet customer needs.” 

Saxena’s work with Caspar AI led him to realize that healthcare, along with other sectors, has a massive staffing shortage, he said

“These problems led us to build TorqueAGI to allow robots to perform tasks that humans cannot,” Saxena said. 

After pushing the boundaries of robotics, Saxena now views the world to be at a pivotal point. “When a technological revolution comes, people often think it is zero-sum for humanity, but this is not true. We are in a place where we will see further democratization of technology. Stanford students get to bring this revolution to the rest of the world” he said.

The post Meet Ashutosh Saxena: The serial entrepreneur proving robots can learn appeared first on The Stanford Daily.