Research Intern — Controllable Video Generation & Diffusion Acceleration

About Company Founded in 2022, XG Tech is driving the future of smart vehicles. Its mission is to empower the digital transformation of automobiles, moving from distributed computing to a centralized, cross-domain platform. XG Tech focuses on the intelligent cockpit—the next frontier of differentiation—while seamlessly integrating advanced driving systems. By reimagining cars as mobile living spaces, XG Tech aligns with the evolving trend of vehicles becoming the “third living space.” Role Summary As a Research Intern, you will explore efficient and controllable video generation for interactive applications. You will investigate how lightweight visual proxies, such as scene layouts, object motion, and camera trajectories, can guide video diffusion models to produce high-quality, temporally consistent content with lower computational cost. Working closely with research mentors, you will contribute to model development, experimentation, and research prototypes, with opportunities to contribute to technical reports and publications. Key Responsibilities • Investigate video generation methods conditioned on structured visual proxies, with control over scene layout, object motion, and camera movement. • Explore diffusion acceleration techniques such as few-step distillation, adaptive sampling, feature caching, and efficient attention. • Research ways to use scene and motion information to reduce redundant computation while preserving visual quality and temporal consistency. • Implement and evaluate models, reproduce relevant research, and conduct controlled experiments. • Support the development of paired proxy–video datasets and training pipelines. • Evaluate generation quality, control accuracy, inference latency, and GPU memory usage. • Document findings and contribute to research discussions, technical reports, and potential publications. How You Will Stand Out • Currently pursuing a PhD in Computer Science, Artificial Intelligence, Computer Vision, or a related field. Exceptional Master’s students with strong relevant research experience are also encouraged to apply. • Strong foundations in deep learning and familiarity with diffusion models, flow matching, or transformer architectures. • Proficiency in Python and PyTorch, with hands-on experience implementing and training deep learning models. • Research or project experience in generative modeling, computer vision, video generation, or model efficiency. • Ability to read research papers, reproduce methods, design experiments, and analyze results independently. • Experience training or fine-tuning image or video generation models. • Familiarity with controllable generation, video-to-video synthesis, world models, or neural rendering. • Exposure to model distillation, feature caching, sparse attention, quantization, or GPU performance optimization. • Publications, open-source contributions, or substantial research projects demonstrating strong research and implementation ability. • Strong curiosity, analytical thinking, clear communication, and a collaborative approach to research.