微信-大模型后训练算法专家
Core
Develop core R&D for LLM inference capabilities including math, logic, and knowledge reasoning to enhance performance in complex scenarios.
Role type
Senior IC large language model post-training algorithm expert
Builds
Optimized LLM inference models for complex reasoning tasks
Domain
Artificial Intelligence / Large Language Models
Deliverable
production ML models
Required skills
LLM inference algorithms, mathematical reasoning, logical reasoning, knowledge reasoning, SFT, DPO, PPO, GRPO, Reward Model design, Transformer architecture, GPT architecture, HuggingFace, Megatron, DeepSpeed, PyTorch
Preferred skills
Independent exploration of frontier technologies, application of research results to business scenarios
Technologies
HuggingFace, Megatron, DeepSpeed, PyTorch
Responsibilities
Develop and optimize algorithms for math, logic, and knowledge reasoning; Track and apply frontier inference technologies to business scenarios