字节跳动Seed启动大模型校招，面向2026.9-2027.8应届生

Core Viewpoint - ByteDance's Seed program has officially launched a recruitment initiative for large model talent, targeting both fresh graduates and current students for internships [1]. Group 1: Recruitment Details - The recruitment is aimed at undergraduate, master's, and doctoral students graduating between September 2026 and August 2027 [5]. - Internships are open to current students graduating in September 2027 and beyond [6]. Group 2: Desired Candidate Traits - The team is looking for candidates with outstanding research and practical skills, strong technical beliefs, high aspirations, curiosity, self-motivation, an open mindset, and a collaborative spirit [8]. Group 3: Technical Research Directions - Foundation Large Models: Exploring the limits of model capabilities and efficiency boundaries [9]. - Visual Intelligence: Focusing on multi-modal pre-training, integration, and unified visual reasoning [10]. - Speech Intelligence: Investigating unified audio generation and understanding, personalized dialogue, and voice agents [11]. - Machine Learning Systems: Concentrating on ultra-large scale training, inference optimization, and hardware-software collaboration [12]. - Large Model Applications: Aiming to optimize the Doubao App experience and build memory and personalization systems [13]. - Embodied Intelligence: Exploring general robotic operation large models to enhance robots' perception, decision-making, and execution capabilities [14]. Group 4: Work Locations - The job locations span multiple cities both domestically and internationally, including Beijing, Shanghai, Shenzhen, Hangzhou, Singapore, San Jose, and Seattle [21].