AI算法工程师（Agentic RL方向）-即梦
Core
Building Agentic RL capabilities for the Jimeng creation Agent to enhance planning, tool use, multi-turn interaction, and task completion in long-chain creative workflows.
Role type
Senior IC Agentic RL Engineer
Builds
Agentic RL systems for image, video, and multimodal editing scenarios
Domain
Generative AI / Multimodal Creative Tools
Deliverable
production ML models
Required skills
Reinforcement Learning, Deep Learning, Large Language Model Post-training, Reward Modeling, Tool Use, Long-horizon Task Planning, PyTorch, Experiment Design
Preferred skills
Agentic RL research, Multimodal RL, Function Call optimization, Failure recovery analysis
Technologies
PyTorch, SFT, DPO, GRPO, PPO, RLHF, RLAIF, Agent Harness, Sandbox
Responsibilities
Design RL tasks, environments, and reward signals for creative scenarios; Construct training data for function calls and task trajectories; Build automated evaluation and replay systems for RL experiments; Analyze failure modes in task planning and execution to drive model and system optimization.
Seniority
Senior, hands-on IC
