微信搜索-AI Infra工程师-大模型后训练与RL方向（北京）（深圳）
Core
Design, develop, and optimize post-training frameworks and RL training for large language models to support efficient and stable training in search scenarios.
Role type
Senior IC machine-learning engineer (LLM post-training & RL)
Builds
High-performance AI search large models
Domain
Search + Large Language Models
Deliverable
production ML models
Required skills
Python, system programming, complex system debugging, distributed training principles, Megatron-LM, DeepSpeed, PyTorch FSDP, PPO, GRPO, DPO, Verl, ROLL, AReal, hardware/software architecture optimization
Preferred skills
Experience deploying and solving problems in RL/distributed computing frameworks, deep understanding of latest LLM structures
Technologies
Verl, Slime, ROLL, AReal, Megatron-LM, DeepSpeed, PyTorch FSDP
Responsibilities
Design and optimize post-training frameworks and RL training; Introduce latest open-source ecosystem technologies; Collaborate with search algorithm teams to implement high-performance AI search models
Seniority
Senior, hands-on IC
