Research Scientist, Body Pose Detection - HYBRID

Research Scientist, Body Pose Detection

Location: Toronto GTA or New York (hybrid) 

Type: Full-time · Openings: 1 

Salary: $150K – $200K (varies by hiring location)

The role

Architect and train proprietary foundation models from scratch for 3D human body tracking and articulated pose estimation. This role targets one of the biggest bottlenecks in embodied AI: full-body kinematics, human locomotion and human-scene interaction. You'll build in-house equivalents to leading 3D body and mesh recovery architectures, trained on a large, continuous stream of proprietary ground-truth human motion data. You'll also lead on new perception problems as the hardware and robotics needs evolve.

What you'll do

• Design, implement and train state-of-the-art networks for 3D human pose estimation, full-body mesh recovery and kinematic tracking


• Scale multi-view and temporal architectures across multi-GPU clusters on massive multi-modal datasets


• Develop novel loss functions that enforce biomechanical constraints, temporal smoothness, balance and physical plausibility


• Build models robust to motion blur, severe self-occlusion and multi-person crowding


• Track bodies through complex spaces, mapping foot-ground contact, joint torques and environmental affordances for downstream humanoid robotics models


• Rapidly prototype models for action segmentation, intent prediction and new hardware sensors


• Feed tracking outputs into physics-aware pipelines that reason about contact, gravity and momentum

What you bring

• Deep expertise in deep learning, 3D computer vision and articulated tracking / human body pose estimation


• Proven experience training large-scale vision models from scratch, beyond inference or fine-tuning existing checkpoints


• Strong grasp of parametric human body models (e.g., SMPL, SMPL-X, GHUM), inverse kinematics and dense mesh estimation


• Mastery of PyTorch and deep learning scaling frameworks


• Experience curating multi-terabyte image and video training datasets


• Comfort in a fast-paced environment where priorities shift with new research or hardware

Strong signals

• First-author publications at CVPR, ICCV, ECCV or NeurIPS on 3D human pose tracking, human-scene interaction, motion capture or human mesh recovery


• Work with large human motion datasets (e.g., AMASS, Human3.6M, EgoBody, PROX)

Please note: Research Scientist roles require highly specific expertise. Apply to only one research role; you may also apply to non-research roles.