Senior AI Engineer

The Senior AI Engineer leads the development of AI-powered features from design through production. The role owns the AI product layer, including LLM integration, prompt design, evaluation, retrieval-augmented generation, agentic pipelines and AI-augmented workflows. The engineer works with Software Engineers to build features that consume and deliver AI capabilities and helps define technical patterns for a 3–4-person delivery pod.

Responsibilities include designing and implementing LLM integrations, structuring prompts systematically, evaluating outputs, iterating from production signals and documenting effective approaches. The engineer develops patterns for streaming responses, function calling, context-window management, fallback handling and evaluation. The role also requires selecting among prompting, RAG, fine-tuning and agentic approaches according to production constraints, contributing production-grade code, and working with a customer-facing agent platform to build, iterate and deploy AI capabilities.

The position contributes code in Ruby/Rails and TypeScript/React and uses Claude Code and GitHub Copilot while critically reviewing AI-generated code. It includes deep code reviews focused on model behavior, prompt design and integration robustness, sharing LLM integration knowledge across pods, and helping teammates develop their AI engineering skills through pairing, documentation and structured feedback.

The engineer independently scopes AI feature work within a 3-week delivery cadence, including experimentation and evaluation cycles, and takes on-call shifts. Operational responsibilities include considering latency, cost at scale, multi-tenant routing and customer data boundaries for AI systems in production. Features are shipped iteratively, with early working versions refined using real usage.

Applicants should have 4–7 years of software engineering experience with significant focus on AI/ML features or LLM-powered product development. Required experience includes prompt engineering, LLM API integration, production AI features, RAG, agentic workflows or function-calling and tool-use patterns. The role also requires full-stack skills in Ruby/Rails and TypeScript/React, fluency with Claude Code, GitHub Copilot and the customer-facing agent platform, systematic LLM evaluation using evals, test cases and production monitoring, operational awareness of production AI systems, and a record of code reviews that develop teammates’ technical capabilities.

This is a full-time role requiring 3 days per week in the Tokyo office. Infrequent travel to Mountain View, California may be needed, usually for one week or less per year.