AI Agent Solution Development Engineer

This engineer designs and leads implementation of AI-agent solutions that use generative AI and large language models. The role covers the full process from upstream design through implementation and operational deployment, working with AI solution architects and customers to turn advanced research technologies into systems used in real business settings.

Responsibilities include developing LLM applications with models such as Claude, GPT and Gemini; building agent solutions with TypeScript frameworks including Mastra, or Python frameworks including LangChain and LangGraph; and applying patterns such as ReAct, tool use, multi-agent architectures and MCP. Use cases include business automation and decision support.

The engineer also designs and implements retrieval-augmented generation systems and knowledge platforms. This includes vector search, hybrid search, metadata filtering, evaluation and tuning. Other duties include integrating external APIs, SaaS products and internal systems, and designing authorization, guardrails and audit logs for workflows suitable for production use.

The role includes building LLMOps pipelines and evaluation platforms, with observability tools such as Langfuse and methods including LLM-as-a-Judge, offline evaluation and A/B testing. Projects are formed on a case-by-case basis and may span finance, manufacturing, satellite data and infrastructure use cases.

Required qualifications include at least three years of professional software development experience; development experience in TypeScript or Python; application development and operations experience in at least one of AWS, Azure or GCP; and experience in at least one of LLM or generative-AI application development, machine-learning or data-processing pipelines, or search and recommendation systems. Team development experience and understanding of Git, code review and CI/CD are required, as is the ability to independently investigate, verify and implement technical solutions.

The posting requires native-level Japanese and English above conversational level, with no resistance to using English. Mastra experience is particularly welcomed, although support is available for candidates with TypeScript implementation experience who have not used Mastra. Experience with Next.js, real-time UI technologies such as Server-Sent Events or WebSocket, LLM frameworks such as LlamaIndex, enterprise platforms such as Amazon Bedrock, Azure OpenAI or Vertex AI, UI/UX design, financial compliance and technical publications or open-source contributions are also welcomed.

This is a full-time, indefinite-term position with a six-month probationary period. Annual salary is 6,000,000 to 12,000,000 JPY, including deemed overtime for 45 hours per month; excess overtime is paid separately. The workplace is in Otemachi, Chiyoda-ku, Tokyo. The standard work arrangement is discretionary labor with eight deemed working hours per day, and the position generally requires office attendance once per week. Remote work is available according to work circumstances and necessary customer visits or field investigations. Benefits include social insurance, commuting expenses up to 40,000 JPY, twice-yearly bonuses, paid leave, refresh leave, career and study support, and expenses for books and necessary equipment.