Build 3D human pose models at Mecka AI using deep learning and articulated tracking to advance embodied AI research. Work model details are not specified.

This Research Scientist role focuses on proprietary foundation models for 3D human body tracking and articulated pose estimation. You will develop networks from scratch, model human-scene interaction, and address complex perception challenges for robotics systems. The work spans multi-view learning, temporal modeling, and physical interaction.

Key Responsibilities:

- Design and train 3D pose estimation networks from scratch
- Develop dense full-body mesh recovery and kinematic tracking
- Scale architectures across multi-GPU clusters and multimodal datasets
- Model occlusion, motion blur,



and multi-person interactions
- Connect tracking outputs with contact and locomotion pipelines

Requirements:

- Deep expertise in deep learning and 3D computer vision
- Experience training large-scale vision models from scratch
- Understanding of SMPL, inverse kinematics, and mesh estimation
- Mastery of PyTorch and scaling frameworks
- Experience curating massive image and video datasets

Apply 3D computer vision, articulated tracking, and model training expertise to Mecka AI’s embodied AI research.

📌 Research Scientist 3D Human Pose (Richmond Hill)
🏢 mecka
📍 Richmond Hill