Machine Learning Engineer

This Machine Learning Engineer role is part of freee’s AI Lab. The engineer will develop machine-learning products used across freee services and investigate new technical fields. The main focus is creating and evaluating products that use large language models and AI Agents, including planning, development, accuracy testing, validation, and improvement of models and agents. The role also includes building and operating a company-wide platform for improving work efficiency with LLMs and a platform for embedding LLM capabilities into products.

Additional responsibilities include developing algorithms for image recognition such as OCR and for anomaly detection. The engineer will build data-processing pipelines, training and evaluation infrastructure, and monitoring pipelines that support continuous product operation and efficient development. The position also involves leading development and evaluation in the LLM and AI Agent field and contributing to data-store capabilities for the AI era. Assignments may change according to business conditions and the individual’s suitability.

Required qualifications include practical understanding and experience in physics, mathematics including statistics, or machine learning; research experience during university may count. Candidates must be able to survey and learn new technologies, have led projects involving multiple stakeholders, have introduced new ideas into products, and have experience developing applications for end users. Desired experience includes developing, evaluating, or operating systems using LLM or AI Agent technologies; interest in Forward Deployed Engineering or Foundation Data Engineering; development and operation of large-scale web services; cost optimization or performance tuning; and development that considers security and internal quality. Experience with image processing, DNN-based OCR, large-scale data processing, or MLOps infrastructure is also welcomed, as are knowledge of accounting or payroll operations, awards in data-science competitions such as Kaggle, and development in or collaboration with domain-specific product teams.

The AI Lab works as one team from service planning through operation and updates its environment as industry trends change. The named development technologies are Python for microservices and the development environment, Ruby on Rails and Go for applications, TensorFlow, PyTorch, and scikit-learn, vLLM, and strands agent sdk. The environment includes a Jupyter Notebook server on Kubeflow running on EKS, container-based development, Visual Studio Code Remote, EKS with GitOps, GitHub Actions, Argo CD, Datadog, Slack, GitHub, MLFlow, Argo Workflows, dbt, langfuse, Looker, Claude Code, and Terraform.

The employment arrangement is permanent full-time employment with a three-month probationary period. The listed workplaces are the Tokyo headquarters in Shinagawa, Tokyo, and the Kansai branch in Miyakojima-ku, Osaka; the workplace may change because of business conditions or office relocation. The schedule uses a specialized discretionary labor system with eight deemed working hours per day. Holidays include Saturdays, Sundays, public holidays, and the year-end and New Year period. Paid leave is granted upon joining, and six paid sick-leave days are provided annually. The stated insurance coverage includes employment insurance, workers’ compensation insurance, employees’ pension insurance, and health insurance. Indoor smoking is prohibited, and online interviews are available.