Agent Harness Engineer

JAPAN AI (Geniee group) is building the Agent Harness — the execution engine that lets AI agents run reliably in production. As an Agent Harness Engineer you will design and implement the execution engine, orchestration, guardrails, memory, and model routing that enable AI agents to operate safely, quickly, and reliably. AI-specific system design covers model routing, context management, and memory infrastructure (long-term memory, working memory, RAG integration), built on an event-driven architecture. Requires 5+ years backend engineering, production Python, and distributed systems design. Stack: Python, Go, TypeScript, React, Next.js, GCP, Docker, Kubernetes, Terraform, Kafka/Pub-Sub, Prometheus/Grafana/OpenTelemetry. Agent frameworks: LangChain, LangGraph, AutoGen; vector databases and RAG systems. Business-level Japanese required; Japan residents only.