anjin.techEvery tech job in Japan
12,021 jobs/3,824 companies
HomeJobsCompaniesReports
About anjin.techContactPrivacy PolicyTerms of Service

© 2026 anjin.tech. All rights reserved.

Browse thousands of tech jobs at Japanese startups and companies. Translated for foreign engineers.

Job listings include English summaries for screening. Always verify details on the original job board before applying.

Jobs→ugo株式会社→[Physical AI x RoboOps] MLOps / Distributed Learning & World Model Evaluation Infrastructure Construction Supporting Domestic AI Robot Development

[Physical AI x RoboOps] MLOps / Distributed Learning & World Model Evaluation Infrastructure Construction Supporting Domestic AI Robot DevelopmentEnglish summary from Japanese

ugo株式会社

✓ Detailed
Salary
700万円〜1100万円
Location
東京都
Job Type
Full-time
Work Style
10:00~19:00
Apply on GREEN

Role summaryEnglish summary from Japanese

English summary for screening — check the original posting before applying.

This role focuses on building and operating the MLOps infrastructure to support robot learning, referred to as 'RoboOps'. You will not be developing models yourself, but rather creating the foundation for smooth model development. The position involves working with robot-specific data and actual hardware.

Must-haves

  • Experience in building and operating ML infrastructure (MLOps)
  • Experience with data pipelines (acquisition, transformation, health check)
  • Experience with large-scale dataset management and storage systems
  • Experience with distributed training infrastructure (e.g., multi-GPU, job scheduling)
  • Experience with evaluation infrastructure for models
  • Proficiency in Python
  • Experience with Linux (Ubuntu 22.04)
  • Experience with Docker

Nice-to-haves

  • Experience with multi-modal robot data (video, behavior, force)
  • Experience with automated data curation for model improvement
  • Experience with SLURM / Kubernetes for GPU cluster workload management
  • Experience with profiling and optimizing GPU utilization, memory usage, and training throughput

About the Company

ugo株式会社

Categories
MarketplaceAI / MLManufacturing
Headquarters
東京都千代田区紀尾井町3-12 紀尾井町ビル11階
Visit Website
Posted8/12/2026
Updated2026/08/12
Sourcegreen
  • Experience with data loader, sharding, and prefetch optimization
  • Experience with world models and their application in evaluation
  • Experience with PyTorch
  • Experience with simulation tools like NVIDIA Isaac Lab / Isaac Sim
  • Experience with data management tools like LeRobotDataset, Safetensors, rosbag / MCAP
  • Experience with communication protocols like gRPC / Protocol Buffers
  • Tech stack

    PythonPyTorchLeRobotRSL RLπ0.5 (openpi)GR00T N1.7Diffusion PolicyACT (Action Chunking Transformer)Flow MatchingNVIDIA Isaac LabIsaac SimLeRobotDatasetSafetensorsrosbagMCAPgRPCProtocol BuffersLinux (Ubuntu 22.04)AWSDockerGitHub ActionsGitSlackNotionClaude CodeCodex

    Work style

    Onsite, Tokyo, Japan. Japanese language proficiency is not explicitly stated as a requirement.