大模型Agent框架工程师 - Seed Model
Core
Building the engineering framework for Large Language Model (LLM) Agents, focusing on planning, tool use, memory, and self-reflection capabilities.
Role type
Senior IC LLM Agent Framework Engineer
Builds
Production-ready Agent frameworks and evaluation benchmarks for applications like Coding Agents and GUI Agents.
Domain
Generative AI / Large Language Models / Agent Systems
Deliverable
production ML models
Required skills
Python, Go, LLMs, Multi-modal LLMs, Agent frameworks, RL training concepts, Benchmark engineering, Distributed systems (Ray, HDFS, Message Queues)
Preferred skills
Experience with SWE-Bench, Tau-bench, RL training pipelines (sampling, reward modeling, on/off-policy training)
Responsibilities
Design and implement core Agent capabilities (Planning/Execution, Tool-Use, Memory, Reflection); Engineer sustainable evaluation benchmarks and metrics; Optimize RL training loops and engineering pipelines; Collaborate with algorithm teams to integrate training and evaluation.
