Cofounder, TLM - GenAI and foundational modeling for map geometry automation in Geo Perception
Founded and technically led a 12-person applied research team responsible for GenAI and scene understanding on PB-scale Geo imagery for map automation. Drove technical roadmap and mentored 4 Staff-level TLs.
Led foundational models (AR, vector diffusion, Gemini Geo post-training) training and inference for 2D/3D building, road, lane map geometry editing, deployed globally and powered >100M automated edits to Google Maps with 95+% accept rate
Awarded 2x Google Tech Impact Award, 4x Geo Tech Impact Awards as TLM
Expertise: VLM/LLM, diffusion models, GenAI map automation, scene understanding, 3D reconstruction, world scale ML system design, JAX, C++, 3-years people management experience
Selected publications:
• Virtual Multi-view Fusion for 3D Semantic Segmentation. Abhijit Kundu, Xiaoqi Yin, Alireza Fathi, David Ross, Brian Brewington, Thomas Funkhouser, Caroline Pantofaru, in ECCV 2020.
• Dilated SpineNet for Semantic Segmentation. Abdullah Rashwan, Xianzhi Du, Xiaoqi Yin, Jing Li, https://arxiv.org/abs/2103.12270
• Learning 3D Semantic Segmentation with only 2D Image Supervision. Kyle Genova, Xiaoqi Yin, Abhijit Kundu, Caroline Pantofaru, Forrester Cole, Avneesh Sud, Brian Brewington, Brian Shucker, Thomas Funkhouser. 3DV 2021 (Oral)
• PanopticNeRF: A Semantic Object-Aware Neural Scene Representation, Abhijit Kundu, Kyle Genova, Xiaoqi Yin, Alireza Fathi, Caroline Pantofaru, Leonidas Guibas, Andrea Tagliasacchi, Frank Dellaert, Thomas Funkhouser. CVPR 2022
Selected impact stories:
• https://lnkd.in/dwQ_qVN3
• https://blog.google/products/maps/google-maps-101-ai-power-new-features-io-2021/
• https://blog.google/products/earth/helping-cities-seed-new-trees-with-tree-canopy-lab/
