大模型基础算法专家-Global Monetization Product and Technology（上海/北京）
Core
Building foundational capabilities for large models and Generative AI, including pre-training, fine-tuning, RLHF, and safety, while developing core Agent capabilities like reasoning, tool use, and long-term memory.
Role type
Senior IC large model and Agentic AI algorithm expert
Builds
Next-generation commercial ecosystem for AIGC in advertising, e-commerce, short video, and live streaming; automated decision-making and long-horizon task agents
Domain
Generative AI, Large Language Models, Reinforcement Learning, Multi-Agent Systems
Deliverable
production ML models
Required skills
Large model pre-training and fine-tuning, Reinforcement Learning, Tool use and skill management, Long-term memory systems, Agentic evaluation and alignment, System-level optimization for low-latency inference, Simulation environment construction
Preferred skills
Experience with Agent Frameworks/Harnesses, Industrial-grade deployment of Agents, Publications in top-tier AI conferences (CVPR, ICML, NeurIPS, etc.) on Agents or RL
Technologies
PyTorch, TensorFlow, ReAct, Plan-and-Solve, Multi-Agent architectures, AgentBench, WebArena, SWE-bench
Responsibilities
Optimize pre-training, post-training fine-tuning, domain knowledge injection, RLHF, and AI safety; Explore and build core Agent capabilities including Agentic RL and autonomous reasoning; Land AIGC technologies in commercial products for content understanding; Develop system-level optimization techniques for efficient training and inference of Agents; Support stable online deployment of Agent solutions with joint training frameworks