World Model Algorithm Engineer

The Amap Machine Learning R&D Department serves as the company's core AI technology engine, focusing on cutting‑edge areas including multimodal large models, video generation and understanding, and image editing and generation. The team is deeply committed to implementing AI technologies, supporting products with hundreds of millions of users, while continuously investing in frontier exploration. We have published multiple papers at top‑tier conferences such as NeurIPS, ICLR, CVPR, ICCV, and ACL, with several works recognized as "Most Influential Papers." With access to massive data and computing resources, we encourage innovation and breakthroughs. We invite you to work alongside leading algorithm experts to jointly define the future of AI! If you aspire to challenge the technical peaks of multimodal and generative AI, achieving breakthroughs at the intersection of video, image, and large models, welcome to join us! Our team's GitHub page: https://github.com/AMAP-ML/
What We Offer
Participate in core AI algorithm development for products serving hundreds of millions of users, witnessing technology directly empower business outcomes
Collaborate with academic leaders and industry experts to continuously expand your technical horizons
Top‑tier computing resources + open innovation culture, supporting frontier exploration and top‑tier conference paper publications
Key Responsibilities
Video Generation: Lead research on frontier video generation technologies, including world models. Possess deep understanding of diffusion and auto‑regressive techniques in AIGC, with research experience in T2I/T2V/I2V, text‑image alignment, and long‑form video generation. Drive the implementation of video generation in Amap's business scenarios (advertising, POI detail pages, etc.).
Video Understanding: Utilize and optimize multimodal large models to perform quality assessment, tagging, dense captioning, and video summarization for user‑uploaded videos. Apply these capabilities to video search, advertising, recommendation, and other business implementations.
World Models: Participate in building world models by leveraging latest video generation techniques—including SFT, RL, DMD+Forcing, and 3D technologies—to construct unified world models that align with human physical laws, and implement them in Amap's business applications.
Stay at the forefront of technology and consolidate technical achievements into top‑tier conference papers and patents.
Job Requirements
Bachelor's degree or above in Computer Science or related fields, with extensive algorithm development experience; practical achievements preferred
5+ years of modeling and analysis experience in NLP, machine learning, data mining, or related fields; experience with big data platform development is a plus
Strong data sensitivity and logical thinking skills, adept at discovering valuable patterns from massive datasets
Passionate about technology, with strong teamwork spirit and ability to thrive under pressure
#J-18808-Ljbffr