Agent: Deep Learning Algorithm Researcher

Agent: Deep Learning Algorithm Researcher

Share

Full-time |Deep Learning Researcher|DeepSeek|Zhejiang · Hangzhou City, Beijing City

Apply for a position

Job description

Job Description:

1. Able to propose innovative ideas and design experimental validation for core issues in large models such as reasoning, generation, and instruction compliance; Exploring new methods and paradigms to enhance model capabilities;

2. Work closely with the data annotation team to design annotation schemes and quality standards; By analyzing data, it identifies shortcomings in model capabilities and constructs targeted training data, forming a closed loop of "data-training-evaluation" capability enhancement;

3. Participate in research on the application of reinforcement learning (RL) in large model alignment and capability enhancement, exploring optimization paths for RLHF/RLAIF, process rewards, preference learning, and other methods;

4. Continuously track cutting-edge advances in the large model field, rapidly reproduce and evaluate the latest industry methods, and promote the implementation of valuable research results.

Job Requirements:

1. Able to independently analyze and propose their own ideas when facing problems, with the ability to drive research from zero to one;

2. Able to quickly turn ideas into actionable prototypes and possess efficient experimental iteration capabilities;

3. Understand the basic concepts and common algorithms of reinforcement learning (such as PPO, DPO, Actor-Critic, etc.); relevant practical experience is preferred;

4. Proficient in Python and C/C++, familiar with deep learning frameworks such as PyTorch;

5. Skilled in cross-team communication, able to collaborate efficiently with data annotation and engineering teams.

Bonus points:

1. Experience in research or projects related to large model pre-training, fine-tuning, or alignment;

2. Published papers at top conferences (NeurIPS, ICML, ICLR, ACL, etc.);

3. Research or practical background in Agent systems, tool calls, and code generation;

4. In-depth use of AI programming tools such as Claude Code and Cursor, with practical understanding of model capability boundaries;

5. Participated in well-known open-source projects or have active GitHub contributions.

Job information

Job title

Agent: Deep Learning Algorithm Researcher

Salary range

-

Nature of the position

Full-time

Types of functions

Deep learning researcher

Department

DeepSeek

Work location

Zhejiang · Hangzhou City, Beijing City