StepFun pr jobs (salary & requirements)

StepFun pr salary range: 30K - 50K, where 100% of positions earn ¥30-50K以上

***  ? 30-50K ***  ? 50K以上

StepFun pr salary range: 30K-50K, the most positions earn 30-50K, based on 3 related positions in the past year, as of 2026-09-19

￥30-50K以上

100% of positions earn

Note: the average salary is analyzed based on job postings published by the company. We recommend reviewing it together with position type, education, region and experience.

StepFun pr salary changes over the years

Note: the data depends on salary samples of online job postings in the corresponding years and does not fully represent the actual situation within the company. For reference only.

Education requirements: 硕士 is the most common

100% 硕士

What education does StepFun require for pr positions? 硕士占比最多，占100%

Salary by education

硕士￥43.3K

StepFun pr salaries by education: 硕士 ¥43.3K.

StepFun pr hiring regions: mainly 北京, 上海

北京￥45.0K 66.7%(2) 򀀩

上海 򀀩

In which cities does StepFun have offices for pr positions? The main distribution is as follows: 北京 accounts for 66.7%. Click this module to view the proportions of other cities, based on job postings in the past year. For reference only.

StepFun pr hiring trends over the years

StepFun pr changes in hiring volume over the years

What does StepFun pr do

Based on related job postings of StepFun in the past year

【26届】LLM pre-train算法工程师
上海-徐汇区 | 硕士以上 | 2026-05-20
30000-35000

【岗位描述】
1、负责Model Architecture的研发、优化和创新，包括Attention、MoE等架构的改进，以及全新架构的探索；
2、负责下一代预训练范式的研究和创新，推动该技术的Scale up；
3、从算法的角度推动大模型训练和推理的低成本化，包括优化器的改进、量化、投机采样等技术；
4、研究通用智能的本质，设计和迭代通用智能的评估和观测方法，从而指导下一代范式的演进；
5、负责大语言基座模型代码、数学、Reasoning能力的提升，系统性的增强模型的能力，实现数据飞轮。

【任职要求】
1、预计毕业时间介于2025年9月至2026年8月的国内外计算机、数学、人工智能等相关专业的博士或优秀研究生；
2、有独立研究能力，曾发表机器学习或相关领域的一作论文；
3、具备扎实的编程基础和优秀的工程能力，能够独立实现相关算法和实验；
4、具有大模型领域的实习或科研经验者优先；
5、具备良好的团队协作能力和沟通能力。
More
【26届校招】LLM pre-train算法研究员
北京-海淀区 | 硕士以上 | 2026-05-27

50000-80000 򀀩

上海 pr salary

Most common salary

30-50K

vs last year

How much does 上海 pr pay? 30-50K is the most common

Discover Great Companies, Discover Great Clients