Nomadic AI Secures $8.4M to Unlock the Value in Autonomous Vehicle Data

For companies building self-driving cars and robots, the greatest challenge isn't collecting data—it's finding the right data. Fleets generate millions of hours of video, most of which sits unused in archives. The tedious work of manually reviewing footage to find specific, rare events slows development to a crawl.

Nomadic AI, founded by Mustafa Bal and Varun Krishnan, is addressing this bottleneck. Their platform applies a suite of vision language models to transform raw video into structured, searchable datasets. This allows engineers to quickly locate specific scenarios—like a vehicle navigating a construction zone or a robot encountering an unexpected obstacle—for both compliance review and model training.

The startup announced an $8.4 million seed round led by TQ Ventures, with participation from Pear VC and Google's Jeff Dean, valuing the company at $50 million. The funding follows a first-place win in Nvidia's GTC pitch contest last month.

"We provide insight on their own footage," Bal explained. "What moves these builders forward is their unique operational data, not random datasets." Clients including Zoox, Mitsubishi Electric, and Zendar use the platform to accelerate development. Zendar's VP of Engineering noted the tool allowed them to scale work faster than outsourcing.

While established labeling firms and Nvidia's open-source Alpamayo models offer similar auto-annotation, Nomadic positions its system as more advanced. Krishnan describes it as an "agentic reasoning system" that uses multiple models to understand actions in context.

Investor Schuster Tanger of TQ Ventures draws a parallel to enterprise software: "The moment an autonomous vehicle company tries to build this internally, they're distracted from what makes them win—the robot itself."

The team, which includes engineers who have all published research, is now developing tools to interpret the physics of maneuvers from video and to integrate non-visual data like lidar. As Bal puts it, managing terabytes of video against massive AI models to extract precise insights remains a formidable technical hurdle—one Nomadic is being paid to solve.

Source: TechCrunch