Our client is an AI-native company established in 2023 as a group company of Geniee, Inc. It builds enterprise AI products with the mission of dramatically expanding human potential through AI technology. With a long-term vision of becoming “the brain of the enterprise,” The client aims to integrate enterprise SaaS systems so AI can safely execute business operations.
The company is focused on turning promising AI workflows into production-grade systems that are reliable, secure, and scalable. Core values reflected in the role include strong engineering rigor, cross-functional collaboration, and a practical focus on quality, safety, and operational excellence.
As an Agent Harness Engineer, you will build the control and execution infrastructure behind production AI agents. This is a foundation software role, not a standard backend API position. Your work will shape the execution engine, orchestration layer, guardrails, memory systems, context management, model routing, and recovery mechanisms that allow AI agents to operate reliably in production.
You will design and implement shared infrastructure used across the products, including the agent execution engine, graph runtime, state machine components, Agent SDK, and workflow orchestration systems. You will also optimize inference pipelines for latency, cost, caching, and reliability, while supporting session management, checkpoints, long-term and working memory, and RAG integration.
The role is highly collaborative, working closely with Research Engineers, Agentic Product Engineers, AI Quality Scientists, Product Managers, Infra, and Data teams. You will also help maintain platform reliability, participate in incident response, and contribute to post-mortems.
Bachelor’s degree or equivalent practical experience in CS, Software Engineering, AI/ML, Math, Physics, or related field
5+ years of backend engineering experience with production Python development
Experience building production systems with LLMs or AI agents
Strong experience with distributed systems, RESTful APIs or gRPC, and [suggestion: production infrastructure]
Japanese fluency and business-level English; Go, Kubernetes, RAG, model routing, guardrails, and MLOps experience are strong pluses
Reviews and bonuses twice per year; stock options available
Full social insurance and full transportation reimbursement
Hybrid work: 3 days in office, 2 days remote; location in Nishi-Shinjuku, Tokyo
Benefits include AI tool support, development/book allowances, language learning support, refresh allowance, and housing allowance
