Research Assistant(s) (carrying the job title of 'AI Research Assistant')

The Center for Mathematical Artificial Intelligence (CMAI) is leading an interdisciplinary research initiative focused on Mathematical Foundations of Autonomous Intelligence Agent. The project aims to develop next-generation, self-motivated Vision-Language-Agent (VLA) models for intelligent operations and predictive maintenance. By integrating cutting-edge multimodal large models, reinforcement learning, and physics-informed neural networks, the project seeks to advance the mathematical interpretability, architectural efficiency, and real-world application of AI systems in high-value industrial settings. The successful candidate will work closely with mathematicians, computer scientists, and industrial partners to translate theoretical models into robust, explainable, and industrially applicable AI systems.

Applicants should have (i) a Bachelor’s or a Master’s degree in Artificial Intelligence, Computer Science, Data Science, Electrical Engineering, Applied Mathematics, or a related discipline; (ii) solid hands-on experience in deep learning framework(s) (PyTorch/TensorFlow/Llama-Factory) and large model fine-tuning and deployment; (iii) practical exposure to multimodal model architectures (e.g., Vision Transformers, VLMs) and reinforcement learning algorithms; (iv) familiarity with training/inference optimization techniques (e.g., distributed training, model pruning, quantization) is a strong plus; (v) basic understanding of mathematical concepts underlying deep learning (optimisation, probability, linear algebra) and interest in model interpretability; and (vi) self-driven, collaborative, and keen to work in an interdisciplinary research environment.

Duties include (a) conducting literature reviews and technical surveys on state-of-the-art vision-language models, multimodal fusion techniques, and self-supervised/self-motivated learning paradigms; (b) assisting in the implementation, fine-tuning, and deployment of large-scale pretrained models (e.g., VLMs, LLMs) for research/industrial perception, diagnostics, and decision-making tasks; (c) developing and optimising training pipelines for multimodal and sequential data, incorporating domain-specific physical constraints; (d) assisting in designing and experimenting with reinforcement learning frameworks for autonomous decision-making and continual learning in dynamic industrial environments; (e) collaborating on enhancing model interpretability through mathematical analysis and visualisation of attention mechanisms, latent representations, and decision pathways; (f) assisting in system architecture optimisation for edge-cloud deployment, including model compression, quantization, and inference acceleration; and (g) preparing technical documentation, experiment reports, and presentations for academic and industrial stakeholders.

Appointment will initially be made on contract basis for up to two years, renewable subject to good performance and mutual agreement.

Applicants should submit a complete curriculum vitae with a full list of publications in the online application.

The University only accepts and considers applications submitted online for the post above via the CUHK career site. For more information and to apply online, please visit http://career.cuhk.edu.hk.