Job Title: Research Engineer, LLM & Agent Systems
Location: Tokyo, Japan
Work Style: Hybrid (3 days in office, 2 days remote)
Employment Type: Full-time
Team Size: Approximately 120 engineers across AI research, platform, voice, and product engineering teams
Company Overview
A rapidly growing AI-native technology company is building next-generation enterprise AI platforms that enable autonomous execution of business workflows through advanced AI agents. Backed by a publicly listed technology group, the company combines startup speed with strong financial resources and has expanded to more than 200 employees and multiple AI products within a few years.
The organization focuses on applied AI research and product development, with the goal of creating enterprise systems that integrate with SaaS platforms and automate complex business operations through large language models, multi-agent systems, and advanced AI infrastructure.
Position Overview
The Research Engineer will lead cutting-edge research and development focused on large language models, AI agents, reasoning systems, memory architectures, retrieval methods, and agent orchestration.
This role sits at the intersection of research and production, applying advanced AI techniques directly to enterprise products used by real-world customers. The position offers the opportunity to solve frontier challenges in agentic AI while driving measurable impact in live production environments.
Key Responsibilities
Agent Research & Development
Design, develop, and evaluate advanced agent architectures, including memory systems, context compression, and inter-agent communication frameworks
Research and implement new approaches to reasoning, planning, retrieval, and tool usage
Develop multimodal AI capabilities spanning text, image, audio, and structured data
Review, reproduce, and extend state-of-the-art academic research
Evaluation & Benchmarking
Design quantitative evaluation frameworks for large-scale agent systems
Build benchmarking methodologies for complex autonomous workflows
Develop synthetic data generation pipelines and evaluation datasets
Support automated model, prompt, and system evaluation throughout the AI lifecycle
Production AI Optimization
Improve model performance, inference quality, and agent reliability
Optimize latency, scalability, and cost through techniques such as quantization, distillation, and caching
Develop and refine training datasets to improve agent performance
Contribute to production deployment and performance tuning initiatives
Cross-Functional Collaboration
Partner with product, engineering, and infrastructure teams to solve complex AI challenges
Support knowledge transfer of research outcomes into production systems
Collaborate with AI platform, evaluation, and product engineering teams
Contribute technical leadership across the AI organization
Research Contribution & Knowledge Sharing
Publish technical findings through papers, technical blogs, or open-source contributions
Participate in collaboration with academic institutions and research communities
Mentor engineers and share best practices across teams
Requirements
Required
Master's degree or PhD in Computer Science, Artificial Intelligence, Machine Learning, Mathematics, Physics, or a related field
Experience developing advanced AI or agent-based systems using large language models
Strong software engineering and machine learning development experience
Deep understanding of LLM and Transformer architectures
Experience with model training and inference using frameworks such as PyTorch or JAX
Strong Python programming skills and ability to write production-quality code
Experience with prompt engineering and LLM-powered applications
Ability to read, reproduce, and improve academic research
Fluent Japanese with business-level English, or business-level English with strong technical communication skills
Preferred
Experience with reinforcement learning for language models
Experience designing multi-agent systems
Research publications in top-tier AI conferences
Experience with RLHF, DPO, or alignment techniques
Experience with multimodal AI models
Background in agent evaluation, AI safety, or AI governance
PhD in a relevant AI or machine learning field
Technology Stack
Python
PyTorch
JAX
Transformers
vLLM
Weights & Biases
GCP
Kubernetes
Docker
TypeScript
React
Next.js
GitHub
Slack
Confluence
Notion
Key Success Metrics
Improvement in benchmark performance across internal and external evaluations
Number of research innovations successfully deployed into production
Reduction in inference latency and operational costs
Quality improvements in agent performance and reliability
Technical publications and knowledge-sharing contributions
Successful technology transfer across engineering teams
What's Attractive About This Opportunity
Opportunity to work on frontier AI challenges including reasoning, memory systems, retrieval, and multi-agent coordination
Direct path from research to production with real-world business impact
Access to modern AI infrastructure, tools, and development resources
Strong collaboration between research and product engineering teams
High-growth environment with significant influence on technical direction
Opportunity to contribute to the future of enterprise AI systems
Benefits
Performance reviews and bonuses twice annually
Stock option program
Hybrid work arrangement
Flexible working hours
Housing allowance
Learning and certification support
Technical book allowance
AI tool subscriptions provided
Wellness and refresh allowance
Language learning support
Modern development equipment and workspace
Interview Process
Application Review
Technical Assignment
Multiple Interview Stages
Reference Check
Offer Presentation