Model Training & Optimization: Design and execute training pipelines for large-scale AIGC models, including data curation, training stability, and performance optimization across computation, communication, and storage layers.
Architecture & System Design: Contribute to the architecture design of video generation models, spanning foundational model training through to post-training alignment and quality enhancement.
Quality Alignment & Evaluation: Build evaluation pipelines and explore alignment strategies to improve model output quality, including preference alignment and reward modeling.
Capability Exploration: Investigate and prototype emerging capabilities such as long-form video generation, multi-modal understanding, and inference acceleration techniques.
Currently pursuing or possesses a Masters or above in Computer Science, AI, or a related field from a reputable university with strong academic credentials/results, expected to graduate by May 2027 and join us by June 2027.
Good publication record at top-tier venues.
Familiarity with video generation architectures and/or diffusion-based generative models.
Demonstrated ability to ship research into working systems.
Strong communication and collaboration skills comfortable working across research and engineering teams.
