Research Engineer - Agent Training Infrastructure (Seed Infra)

Beijing ByteDance Technology Co Ltd

Seattle, WA

JOB DETAILS
SKILLS
Artificial Intelligence (AI), Benchmarking, Computer Science, Cross-Functional, Debugging Skills, Distributed Computing, Environmental Management, GPU (Graphics Processing Unit), Machine Learning, Mentoring, Open Source, Performance Management, Performance Tuning/Optimization, Prototyping, Publications, Python Programming/Scripting Language, Reinforcement Learning
LOCATION
Seattle, WA
POSTED
30+ days ago

About the Team

The Seed Infrastructures team oversees the distributed training, reinforcement learning framework, high-performance inference, and heterogeneous hardware compilation technologies for AI foundation models.

Responsibilities

  • Design, implement, and maintain agent execution environments and runtime frameworks for multi-agent training at scale

  • Build and optimize infrastructure for RLHF pipelines, reward modeling, and distributed RL training

  • Manage and orchestrate many-agent parallel execution, including environment simulations and environment managers

  • Collaborate closely with research teams to support the LLM training pipeline: training ? SFT ? RLHF ? evaluation ? serving

  • Ensure high-performance, scalable, and fault-tolerant distributed systems for agent frameworks

  • Develop tools and libraries to monitor, debug, and benchmark agent training and inference

  • Translate research prototypes into production-ready infrastructure that can support large-scale AI experiments

Minimum Qualifications

  • M.S. or Ph.D. in Computer Science, Machine Learning, or a related field

  • Strong experience with Python and distributed systems frameworks (e.g., Ray)

  • Hands-on experience building agent infrastructure: execution environment, runtime, or environment manager

  • Experience managing parallel multi-agent execution, including simulations and environment orchestration

  • Familiarity with the LLM pipeline (training ? SFT ? RLHF ? evaluation ? serving)

  • Proven ability to design and maintain high-performance, scalable, and robust distributed AI systems

Preferred Qualifications

  • Experience building or contributing to RLHF pipelines, reward modeling infrastructure, or RL training infrastructure

  • Strong understanding of multi-agent reinforcement learning and agent orchestration at scale.

  • Familiarity with GPU clusters, distributed training strategies, and performance optimization

  • Publications or open-source contributions in agent systems, distributed RL, or AI infrastructure.

  • Experience mentoring engineers and collaborating in cross-functional research and engineering teams

About the Company

B

Beijing ByteDance Technology Co Ltd