Xin Tong, a Distinguished Scientist at Meta who recently completed postdoctoral research at MIT CSAIL, has joined Meshy as Chief Scientist and Co-founder. The move pairs a leading figure in AI research with the Tsinghua University-affiliated company, aiming to advance generative AI for 3D model construction at a moment when the industry is shifting toward AI-driven content creation.
Tong conducted his postdoctoral work under MIT professors Joshua Tenenbaum and Song Chunxiao. At Meshy, he will lead the company's long-term research efforts in AI and 3D modeling. The collaboration draws on research strengths from both Tsinghua University and MIT, with support from the Chinese Academy of Sciences and the Ministry of Science and Technology.
What the research team has built
Tong and his team have developed a multimodal 3D reconstruction platform and an AI-powered 3D design tool. The systems generate highly detailed 3D models from text prompts or 2D images. According to published research, the technology reduces both production costs and time, bypassing steps that traditionally required manual modeling.
Meshy's platform already supports the generation of 3D models with facial expressions, anatomical structures, and skin textures. It also reproduces text and patterns with precision. The company reports its user base has passed 10 million.
An industry in motion
The appointment arrives during a period of rapid change in 3D content creation. Earlier this year, Google launched Genie 3, a platform that generates explorable 3D environments from text. In the same window, Unity's stock dropped 24% and Roblox's fell 27%, signaling investor recalibration around traditional game engines and user-generated content platforms.
Industry analysts have pointed to these shifts as early indicators that video models could eventually replace conventional game engines. Meshy's latest model series, Meshy-7, has made progress on geometric alignment, addressing a specific bottleneck in AI-driven 3D modeling.
Why this matters for science and research professionals
For researchers and engineers working in computer vision, graphics, or generative modeling, Tong's move signals where high-level AI talent is concentrating. The intersection of multimodal learning and 3D reconstruction is producing tools that compress workflows from days to minutes. Professionals who track model architecture decisions, training data requirements, and evaluation benchmarks will find Meshy's published advances in geometric alignment worth close examination. The collaboration also highlights how academic labs - particularly MIT CSAIL and Tsinghua - continue to feed directly into product-focused AI startups, shaping the talent pipeline and research agenda simultaneously.
Your membership also unlocks:
