Tiancai Wang
Tiancai Wang is a co-founder at Dexmal and an affiliated researcher at Nanjing University, working at the intersection of embodied AI and robotics. He previously served as a senior researcher at Megvii Technology, where he built a body of work in computer vision exceeding 7,400 citations. Wang is best known as a corresponding author on the Realtime-VLA FLASH paper (arXiv 2605.13778), which introduces a speculative inference framework that achieves roughly three times faster inference for diffusion-based vision-language-action models without retraining the main model, enabling sub-200ms end-to-end latency for real-time robotic manipulation.
“FLASH is a software-level framework that makes the same model run 3x faster at inference time, with almost no degradation in task success — no retraining the main model, no new hardware.”
Source→“Tiancai Wang — Dexmal / Nanjing University; Corresponding author. Positioned at the intersection of academic research and Dexmal's commercial deployment focus on latency-aware VLA systems.”
Source→AI-extracted from podcast / newsletter / paper summaries. May contain errors.