Job Description

SuperSparks - 校招 - 智能座舱大语言模型算法工程师上海、北京校招正式数字技术 - 算法博士及以上Super Sparks 招聘计划职位 ID：A147661职位描述1. 针对车载人机交互AGI系统，实现相关算法与功能的设计和实现
2. 围绕C端用户每日长时间交互诉求，实现垂域agent算法架构、数据仿真和收集和相关垂域模型训练
3. 基于RL、数据模拟、context & tool scaling等技术，研究用户意图理解、个性化交互、多轮交互和记忆等长周期任务相关的训练、数据构造及效果分析等
4. 协助真实数据问题挖掘、分析及解决方案制定等，有效解决实际系统中的核心高频和长尾问题
1. Design and implement algorithms and functions for in-vehicle human-machine interaction AGI systems.
2. Develop vertical domain agent algorithm architectures, data simulation, data collection, and training of relevant vertical domain models, focusing on meeting C-end users' daily long-duration interaction demands.
3. Leverage technologies such as Reinforcement Learning (RL), data simulation, and context & tool scaling to research training methodologies, data construction, and effect analysis for long-cycle tasks including user intent understanding, personalized interaction, multi-turn interaction, and memory.
4. Assist in identifying and analyzing real-world data issues, formulating solutions, and effectively addressing core high-frequency and long-tail problems in practical systems.职位要求1. 毕业于国内外顶尖高校或者研究院所，博士及以上学历，计算机、自动化、信号处理等相关专业
2. 有大模型(如Qwen、DeepSeek、GLM等)的相关训练经验者优先，深入理解大模型预训练、后训练、RL训练相关流程
3. 有高水平论文发表、开源社区贡献或相关竞赛名次优先
1. Graduated from top domestic and international universities or research institutions, with a PhD or higher degree in Computer Science, Automation, Signal Processing, or related fields.
2. Preferred: Prior experience in training large models (e.g., LLaMA, Qwen, DeepSeek, GLM), with in-depth understanding of large model pretraining, post-training, and RL training processes.
3. Preferred: Publications in top-tier conferences/journals, contributions to open-source communities, or rankings in relevant competitions. 投递