Synthetic Data, Evaluation, and the Future of Simulation for Physical AI w/ Charles Wong + Aravind Kandiah

If you want to understand what it takes to get robots out of the lab and into production, this one’s for you.Today, I’m joined by the cofounders of Bifrost: Charles Wong (CEO) and Aravind Kandia (CTO).Bifrost is one of the leaders in synthetic data generation and simulation for evaluation. They’ve built some of the most impressive tools I’ve seen to date for helping robotics developers train, test, and deploy autonomy in the real world.In this episode, we discuss:* what most teams still get wrong with simulation* why real-world data alone is not enough* why evaluation is becoming the real bottleneck* how the best teams are closing the infamous sim-to-real gap* where simulation will be in 5-10 yearsThese guys are two of the sharpest founders building in robotics right now — and they’ve been at it long before the physical AI hype train. I think you’ll find their insights super valuable and rich with signals from the frontlines.With that, enjoy this episode with Charles and Aravind. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.dreammachines.ai