Mid-Level AI Data Platform Engineer

This full-time, mid-career engineering role in Tokyo develops the foundation that enables all freee employees to reach reliable business data through conversations with AI agents. The engineer designs semantic layers and context infrastructure based on user demand, delivers analysis agents and remote MCP services, and measures and improves answer quality through logging and evaluation. Responsibilities also include data product management, facilitation, and adoption across the company. Assigned duties may change according to business conditions and the candidate’s suitability.

The work supports freee’s integrated cloud software for small businesses and individuals, including accounting, human resources and labor management, sales management, tax filing, and electronic signatures. The data platform spans multiple products and business areas, with up to tens of thousands of tables, more than one million monthly queries, dozens of microservice databases, and data from SaaS and financial businesses. Thousands of users access the platform each month. The initiative aims to make analysis, insights, and initial drafts available through natural-language questions, including for people who are not data-analysis specialists. Reliable AI-ready data requires machine-readable definitions for metrics, business terms, and context.

Required qualifications include backend development with Python and production operation on Google Cloud or an equivalent public cloud. Candidates must have experience developing LLM applications such as RAG systems, agents, MCP servers, or evaluation systems. This may include high-quality personal development or open-source work when the design intent can be explained. Understanding of SQL using a data warehouse such as BigQuery and dimensional modeling is required, as is development experience with Docker and CI/CD tools such as GitHub Actions.

Preferred experience includes practical work with semantic layers such as LookML; LangGraph, LangChain, or similar frameworks; LLM evaluations and evaluation-driven development; measuring and improving application quality using logging and metrics such as accuracy and resolution rate; and designing or operating remote MCP servers, including OAuth authentication and authorization. Product management experience, including internal or data products, is also welcomed. Experience or interest in data governance, access control, metadata management, data catalogs, ontologies, OKF, data mesh, hub-and-spoke models, BigQuery Sharing, policy tags, and B2B SaaS domains such as accounting, human resources, and sales data is relevant.

The posting seeks someone who understands and identifies with freee’s stated value criteria, prioritizes a model that is used over one that is merely correct, explains quality through measurements, selects technology based on operating cost and organizational circumstances rather than trends, and documents definitions and the reasons behind decisions. Online interviews are available.