Research Engineer, Embodied AI & Robotics, IAIC

We are seeking a motivated research engineer to develop and deploy advanced computational systems in embodied artificial intelligence and robotics. The candidate will work closely with research scientists to implement, optimize and deploy AI models for multiple domains, such as robotics and healthcare, with a focus on human-AI teaming/collaboration. Examples of R&D areas include multimodal human motion analysis, human intent prediction, and human digital twins.

Responsibilities

• Develop and implement deep learning models for human motion analysis, human–object interaction, and force-aware motion modelling.

• Build and maintain data processing pipelines for multimodal human motion datasets.

• Train, fine-tune, and evaluate neural networks for motion prediction, interaction understanding, and force estimation.

• Collaborate with researchers, engineers, domain experts, and external partners on applied research projects.

• Document technical work and contribute to publications, reports, demonstrations, and project deliverables.

• Optimizing computational performance using GPU acceleration and high-performance computing platforms.

• Supporting the development of research prototypes and experimental platforms.

Requirements

• Bachelor's or Master's degree in Computer Science, Computer Engineering, Artificial Intelligence, Data Science, or related disciplines.

• Strong programming skills in Python and PyTorch.

• Good understanding of machine learning, deep learning, and computer vision.

• Ability to work independently on research/engineering problems and communicate technical ideas clearly.

• Experience with one or more of the following is an advantage:

  • Work with RGB/RGBD data, 3D human pose (SMPL), hand pose (MANO), object pose, mesh representations, and force-related signals

  • 3D Human and hand pose estimation, and motion forecasting

  • Generative models for motion synthesis

  • Multimodal learning

  • Biomechanical or force-related motion analysis via the OpenSim library