Research Engineer, LLM Modeling
Salary (published): ¥857,143–¥1,428,571 / month incl. 45h fixed overtime — approximately ¥10.29M–¥17.14M / year, plus stock options; negotiable. Languages: Business-level English or fluent Japanese.
About the role
Research-grade engineering on frontier agent systems: reasoning quality, long-term memory, and multi-agent coordination, with a direct path from research into JAPAN AI’s product teams.
Responsibilities
- Research cutting-edge agent systems
- Develop benchmarks and optimize inference
- Transfer innovations to product teams
- Work on frontier challenges: reasoning quality, long-term memory, multi-agent coordination
Qualifications
- Master’s or PhD in CS, AI, ML, or a related field
- Deep experience building agentic systems with LLMs
- Strong Python implementation skills
- Understanding of Transformer architectures
Nice to have
- Large-scale reinforcement learning
- Multi-agent systems
- Top-tier conference publications
- Alignment techniques (RLHF / DPO)
- Multimodal models
Tech stack
Python, PyTorch, JAX, Transformer architectures, LLM systems, reinforcement learning.
External listing — the salary range is stated on the official posting (verified live 2026-07-30). The Apply button takes you to JAPAN AI’s official HRMOS page.
Working conditions
- Employer
- JAPAN AI
- Employer contact
- https://hrmos.co/pages/geniee
- Workplace
- Tokyo (Nishi-Shinjuku) · Hybrid (3 office / 2 remote)
- Type
- Full-time
- Salary
- ¥10.29M–¥17.14M / year (approx.)
These working conditions are summarized from the employer's official posting, where the full and authoritative conditions are published. Please confirm the final terms directly with the employer before accepting any offer.