智能体-强化学习算法研究员-CodeBuddy/WorkBuddy
Core
Researching and designing Agentic Workflow and Agentic Memory solutions to solve problems in the code domain, with a focus on Reinforcement Learning (RL) that outperforms Supervised Fine-Tuning (SFT).
Role type
Senior Research Scientist (Agentic RL & Engineering)
Builds
Scalable Agentic Workflow solutions for code assistance
Domain
Artificial Intelligence / Reinforcement Learning / Software Engineering
Deliverable
production ML models
Required skills
Reinforcement Learning, Agentic Workflow design, Agentic Memory architecture, Deep Learning optimization, LLM inference optimization, Python, C/C++, Golang, Java, JavaScript, TypeScript
Preferred skills
Publications in top-tier conferences (ACL, EMNLP, NeurIPS, ICML, ICLR)
Responsibilities
Designing memory access logic for agents; Adjusting training corpora and objectives based on user data and agent implementation details; Designing frontend and backend architectures to optimize LLM inference costs; Collaborating with agent developers to align model training with agent design.
Seniority
Senior, hands-on IC
