Senior AI Engineer - LLM Adoption Department (LLMAD)

The AI & Data Division (AIDD) creates powerful, customer-focused search, recommendation, data science, advertising, marketing, price, and inventory optimization solutions to a variety of businesses in the Rakuten group. Our goal is to drive innovation by developing new products and capabilities that deliver significant impact over longer timeframes using AI.

The LLMAD team drives adoption of Rakuten AI and open-source LLMs hosted within our own environment. We're hiring AI Engineers who can be embedded directly inside business units to drive real adoption of in-house large language models in place of third-party alternatives.

As a Senior AI Engineer you will own a business unit engagement end to end. You sit inside their engineering team and build with them: designing the agent, selecting and evaluating the model, writing the prompts and tool-use logic, and getting it live in production on our own infrastructure.

You are expected to operate independently - taking a vague business ask, turning it into a concrete and evaluable LLM task, and being accountable for it shipping. What you build also becomes reusable: the cookbooks, recipes, and reference implementations the rest of the team and other business units start from.

Own a business unit engagement end to end - embed in their engineering team, design the agent (task decomposition, tool and function calling, multi-step orchestration, state, retrieval, guardrails), and write production code in their codebase.

Take agents from prototype to production and own what that requires: latency, cost, failure modes, and monitoring - including the judgment to say early when an agent is the wrong answer and a prompt, a fine-tune, or plain software is not.

Select the right in-house or open-source model for each task and prove it with task-level evaluation rather than benchmark scores, building the eval sets and harnesses together with the business unit.

Run migrations from third-party APIs to our in-house models - parity testing, prompt porting, staged cutover - and feed the gaps you find back to our model and platform teams.

Own prompt and context engineering as a discipline: structured output, tool schemas, retrieval strategy, and caching, versioned and measured against evals; debug quality problems down to whether the fault is the prompt, the retrieval, the model, or the data.

Turn the work into cookbooks, recipes, and reference implementations other business units can lift directly, and raise the bar around you through code review, mentorship, and hands-on enablement for business unit engineers.

You will not be training or fine-tuning foundation models - that sits with our model research and training teams.

You will not be building the LLM platform itself - that sits with our platform engineering teams.

You will be embedded in a business unit, writing agent and integration code inside their product, and accountable for it shipping on our models.