Research Scientist: 3D/4D Reconstruction/Generation

About the Team The PICO MR team is dedicated to developing core environmental perception algorithms for the PICO system. The team focuses on three main areas: visual localization, scene perception, and 3D reconstruction. Its work in MR scenarios includes research, exploration, and deployment of technologies such as multimodal scene perception, AIGC scene generation, depth estimation (Mono/Stereo/MVS), 3D scene reconstruction (3DGS, NeRF, etc.), and lighting estimation. The team boasts strong technical capabilities, with most members holding master’s or doctoral degrees from renowned universities (e.g., Tsinghua University, Peking University, Zhejiang University, Nanyang Technological University, Hong Kong University of Science and Technology). It has published numerous papers in top conferences/journals like CVPR, ECCV, ICCV, SIGGRAPH, and TPAMI. The team has won multiple awards: CVPR 2025 OpenSUN3D Double Track Champion, CVPR 2025 Cross-device Localization Challenge Champion, CVPR 2024 Monocular Depth Estimation Challenge Champion, CVPR 2024 OpenSUN3D 3D Functionality Grounding Track Champion, and ICCV 2023 OpenSUN3D 3D Object Instance Search Track Champion. Responsibilities 1. Participate in R&D for 3D/4D algorithms and foundation models, and promote their deployment in MR application scenarios. 2. Develop and optimize large 3D/4D generative models (e.g., data curation, pre-training, post-training, evaluation, acceleration and deployment). Minimum Qualifications - Master’s degree or above in computer science or related fields; 3+ years of algorithm research or industry experience. - Excellent programming skills; proficient in Python/C/C++ and common deep learning frameworks. - Solid foundation in AIGC/3DV/CG; project experience in reconstruction or generation. - Proficiency in 3D representation/rendering (Gaussian Splatting, NeRF, meshes, point clouds); Preferred Qualifications - Thorough understanding of the latest advances in reconstruction and generation. Preferred experience in any of: - 3D/4D reconstruction (feed-forward GS/optimization-based) - 3D/4D generation (human/scene/novel view synthesis) - Image/video generation & understanding (video depth estimation/tracking/understanding) - Publications in CV/CG/ML/AI conferences or journals (e.g., CVPR, ICCV, ECCV, SIGGRAPH, TOG, TVCG, NeurIPS, ICML, ICLR) focusing on computer vision/computer graphics. - Hands-on experience in building large model training pipelines and curating datasets. - Familiarity with MR/AR/VR technologies is a strong plus. - Award winners of competitions such as ACM-ICPC, NOI/IOI, TopCoder, or AI/CV contests are preferred.