Ali Agha

Founder & CEO FieldAI

Last Reviewed: September 23, 2026

Executive Summary

Ali Agha is the founder and chief executive of FieldAI, an Irvine, California company that builds AI “brains” for robots working in places nobody has mapped: construction sites, energy and industrial facilities, and city streets. Its Field Foundation Models are designed to run on different kinds of robot and to navigate without maps, GPS or predefined routes. In August 2025 the company said it had raised $405 million across two consecutive rounds, from backers including Bezos Expeditions, NVIDIA’s NVentures, Khosla Ventures and Temasek. Agha, who publishes as Ali-akbar Agha-mohammadi, comes to the job from research. He spent seven years at NASA’s Jet Propulsion Laboratory, where he led the JPL-based team that won the Urban phase of the DARPA Subterranean Challenge. FieldAI’s pitch rests on that background: autonomy built for risk and unpredictability from the start, rather than language models retrofitted onto machines.

Career Highlights

• Founder and CEO of FieldAI, which announced $405 million raised across two rounds in August 2025.
• Signed partnerships with Boston Dynamics (March 2026), pairing its Spot robot with FieldAI’s models on construction sites, and with Caterpillar (September 2026).
• Led Team CoSTAR, a JPL, MIT, Caltech, KAIST and LTU collaboration, to first place in the 2020 Urban phase of the DARPA Subterranean Challenge.
• Principal Investigator at JPL on DARPA RACER, NASA’s Autonomous Mars Cave Exploration, and coordinated autonomy for a prototype Mars helicopter and rover.
• Research Staff Engineer at Qualcomm Research, 2015 to 2016, and postdoctoral researcher at MIT, 2013 to 2015.
• PhD in computer science and engineering from Texas A&M University.

Professional Journey

Agha trained in electrical engineering with a focus on control systems, earning bachelor’s and master’s degrees in 2005 and 2008. He then completed a PhD in computer science and engineering at Texas A&M University between 2009 and 2013, and was also a graduate research assistant in aerospace engineering there. He spent 2013 to 2015 as a postdoctoral researcher at MIT and a year at Qualcomm Research as a research staff engineer working on robotics.

He joined NASA’s Jet Propulsion Laboratory, run by Caltech, in 2016 as a robotics technologist and later a group leader. Over seven years there he was Principal Investigator on some of the lab’s most visible autonomy programs: the DARPA Subterranean Challenge, DARPA’s RACER off-road autonomy program, NASA’s work on autonomous exploration of Martian caves, and safety-critical coordinated autonomy for a prototype Mars helicopter and rover. For the Subterranean Challenge he led Team CoSTAR, drawn from JPL, MIT, Caltech, KAIST and LTU, which won the competition’s Urban phase in 2020 with robots that explored complex, unknown environments on their own.

FieldAI grew out of that work. The company argues that robots in the physical world need architectures that are aware of risk from the ground up, not large language and vision models bolted on afterwards. Its models are “embodiment-agnostic,” meaning they can run on legged robots, wheeled vehicles and other machines. In August 2025 FieldAI said its robots were in daily use across hundreds of industrial sites in the US, Europe and Japan, in construction, energy, manufacturing, urban delivery and inspection. The same announcement disclosed $405 million raised across two rounds, from investors including Bezos Expeditions, BHP Ventures, Canaan Partners, Emerson Collective, Intel Capital, Khosla Ventures, NVentures, Prysm and Temasek, with earlier backers Gates Frontier and Samsung returning.

In 2026 Agha has moved FieldAI into partnerships with established equipment makers. In March the company teamed up with Boston Dynamics to run its models on the Spot robot for construction work, and it announced a collaboration with NVIDIA on industrial AI. In September it announced a collaboration with Caterpillar, combining Caterpillar’s operational data with FieldAI’s robot foundation models for jobsites and manufacturing plants. He continues to publish research, and his personal site lists recent papers on vision-language models for robotics, humanoid control and mapping in extreme environments.

Education

Texas A&M University

PhD, Computer Science and Engineering, 2009-2013

He also holds bachelor’s (2005) and master’s (2008) degrees in electrical engineering with a focus on control systems; the institution is not named in any source consulted.