AI Multi-modal & World Model Scientist

Responsibilities

• Design and implement Multi-modal Large Models (VLM/LLM) and World Models, including model selection, SFT, and RL tuning, building prediction and reasoning capabilities for environmental dynamics.
• Participate in Agent application research, building next-generation end-to-end multi-agent systems for autonomous planning and execution of complex tasks.
• Lead engineering design and implementation of NLP/CV/LLM algorithms, optimizing performance for industrial scenarios to ensure efficient deployment on edge or cloud.
• Explore and practice Code Agent assisted development workflows, leveraging AI toolchains to improve efficiency and quality in code generation, debugging, and system architecture.

Requirements

• Bachelor's degree or above in Computer Science, EE, Mathematics, AI, or related fields; Strong engineering capabilities may override degree requirements.
• Deep understanding of mainstream LLM/VLM/VLA principles and architectures; In-depth knowledge of World Models (e.g., environment representation, dynamic prediction, policy generation).
• Hands-on experience in full-stage LLM training (including Pre-training, SFT, RLHF/DPO), with proven track records in training from scratch or deeply fine-tuning mainstream open-source models.
• Proficient in Python/C++; Experience using Code Agents (e.g., advanced Cursor/Copilot workflows, custom Agents) for efficient development is highly preferred.
• Publications in top conferences (CVPR, NeurIPS, ICLR, ICML, ACL) are a plus; Experience in Embodied AI or industrial simulation scenarios is a plus.

What We Offer

• Competitive salary package plus annual review and project bonuses for high performers.
• Access to cutting-edge World Model and Embodied AI scenarios with ample computing resources (A100/H800 clusters) and data support
• Collaborate with top Hong Kong university research teams (CUHK/HKUST/HKU/HKBU) and world-class AI experts, participate in international joint projects, publish at top conferences, and expand academic influence.
• Result-oriented culture supporting technical exploration and innovation, with dual career tracks: "Technical Expert" and "Technical Management".
• Provide HK Visa support