Multimodal Affective Understanding Model Engineer
Algorithms · Full time · Shenzhen (work arrangement discussed for the role)
Compensation discussed individually
Role overview
Build multimodal affective-understanding models from facial, body-language, and related signals for edge deployment.
Responsibilities
- Lead the design, training, evaluation, and tuning of multimodal perception models.
- Develop information-detection and affect-recognition capabilities around facial and body-language signals.
- Optimize inference speed and resource use for stable real-time execution at the edge.
- Integrate and deploy models at the edge, establishing performance baselines, validation flows, and monitoring.
- Evolve multimodal affective-understanding capabilities toward product delivery.
Requirements
- Graduate degree in computer vision, deep learning, AI, or a related field, or a bachelor degree plus two years of relevant experience.
- Strong computer-vision fundamentals across object detection, image classification, keypoint detection, or related areas.
- Practical experience training, tuning, and deploying models with PyTorch.
- Strong problem analysis and experimental iteration skills with the ability to carry algorithms into engineering delivery.
- Ability to collaborate with software, hardware, and product teams on real-world validation.
Preferred qualifications
- Research, project, or delivery experience in facial recognition, affective computing, or multimodal perception.
- Familiarity with ONNX, TensorRT, mobile deployment, or edge NPU optimization, with practical examples.
- Algorithm experience for robotics scenarios or participation in RoboMaster, RoboCon, or comparable competitions.
Information notice
Submit only information relevant to a careers conversation and do not include unrelated sensitive personal information. Submitted information may be used to review and reply to your inquiry.