An Phan builds the data foundations beneath physical AI systems. Working at Hippo Harvest, he designs architectures for sensor-rich, autonomous platforms operating in the real world. He tackles environments where signals are incomplete, conditions change, and lost telemetry means you cannot simply recompute reality.
Rather than chasing novelty, An focuses on reliability and reproducibility across on-prem and cloud setups. His work ensures that pipelines for robotics and physical domains capture irreversible edge data reliably, maintaining an accurate and unbroken history between models and real-world decisions.
Videos
You Can’t Rerun Sunlight: Designing ML Data Architectures for Physical AI
An Phan · Hippo Harvest
World Congress North America · 2026