Agent Harness Engineer

JAPAN AI, Inc. (a subsidiary of Geniee, Inc., listed on the TSE Growth Market) builds JAPAN AI STUDIO, described as "the brain of the enterprise" that autonomously executes hundreds of workflows across customer deployments.

The Agent Harness is the control layer that wraps the model and manages session state, checkpoints, guardrails, context injection, and tool execution. It is what transforms an agent from "works in a demo" to "trusted in production." As an Agent Harness Engineer you will design and implement the Agent Harness (execution engine, orchestration, guardrails, memory, and model routing) that enables AI agents to operate safely, quickly, and reliably.

How this differs from a Backend Engineer: a Backend Engineer manages stateless request/response with authentication controls. An Agent Harness Engineer manages agent session state, model routing, memory infrastructure, and policy execution. It is a new domain at the intersection of AI/ML and infrastructure.

Responsibilities:

• Design and implement the agent execution engine (Graph Runtime / State Machine) with deep understanding of LLM / AI agent operating principles
• Own AI-specific infrastructure including model routing and memory management
• Build an Agent SDK used by ~120 in-house engineers
• Implement model routing across multiple LLM providers, RAG integration, and inference optimization (latency, cost, caching)
• Develop guardrail / policy execution engines to control agent behavior
• Handle workflow orchestration and load balancing
• Collaborate with Research Engineers on production integration

Requirements:

• Bachelor's degree in CS, Software Engineering, AI/ML, Mathematics, Physics, or equivalent
• 5+ years backend engineering experience
• Production Python development
• Experience with LLM / AI agent production systems
• Distributed systems design and implementation experience
• RESTful APIs / gRPC expertise
• Business-level English OR fluent Japanese

Strongly preferred:

• Agent framework / harness implementation experience
• Cloud production operations (AWS/GCP/Azure)
• RAG systems, vector databases, and memory architecture knowledge
• Model routing / inference optimization
• Go for foundation software, Kubernetes expertise
• Event-driven architecture (Kafka/RabbitMQ)
• Safety guardrails and AI observability

Strong candidates have experience with agent frameworks such as LangChain, LangGraph, and AutoGen.

Tech stack: Python, Go, TypeScript/React/Next.js, GCP (Kubernetes/containers), Docker, Terraform, Kafka, Pub/Sub, Prometheus, Grafana, OpenTelemetry.

Compensation: Monthly ¥857,143–¥1,428,571 (includes 45 hours fixed overtime; overtime beyond 45 hours paid separately). Stock options available. Reviews and bonuses twice yearly. Negotiable based on experience. Company-covered AI tool costs (Claude, ChatGPT, Cursor), development tool/book allowances, language learning support, refresh and housing allowances.

Location: Tokyo (Shinjuku, Sumitomo Realty New Shinjuku Oak Tower). Hybrid: 3 days in office, 2 days remote. Hours 10:00–19:00 (core hours negotiable).