応募資格

Experience independently defining what problem should be solved in a complex and ambiguous user domain, and turning it into a 0→1 product.
Experience launching a consumer-facing product to real users and continuously improving it based on usage data and user feedback.
A track record of not only solving difficult problems, but also creating significant changes in user behavior or experience as a result.
Experience defining what a good outcome means for a problem with no single correct answer, and designing the evaluation methods and metrics for it.
Experience designing systems that adapt the feedback or experience they provide based on the user’s state and past behavior.
Experience working deeply with domain experts and translating their knowledge and decision criteria into products or systems.
Experience determining the overall technical architecture and development direction of a product across Backend, AI, and Product.
Experience defining the problem, evaluation criteria, and Technical Direction from scratch for unprecedented technical challenges, and codifying them in a form other engineers can reference.
Experience influencing product strategy and business priorities through technical decision-making.

Experience developing, operating, and improving a consumer-facing product used by millions of users.
Experience building systems that infer latent information—such as intent, understanding, skill, or state—from user behavior and observational data.
Expertise in sports coaching, coaching for musical performance or voice training, Clinical Education, or other forms of learning and mastery support that go beyond transferring knowledge and enable practitioners to observe changing situations and make their own judgments and actions in the moment.
Experience structuring experts’ tacit knowledge and translating it into reusable rules, representations, or AI systems.
Experience tracking user behavior and growth across multiple sessions and over extended periods, and using those insights to improve a product.
Experience developing products that use data to understand the state of the real world in domains such as Video, Sensor, Healthcare, Sports, Manufacturing, or Robotics.

募集要項最終確認 10/5

職種: AIエンジニア
雇用形態: 記載なし
役職: Senior Principal Level
給与: 記載なし
給与の詳細: There is an opportunity for salary revision once a year based on the evaluation results of the first and second half of each fiscal year.
勤務地: 〒153-0044 2-22-44, Ohashi, Meguro-ward, Tokyo
働き方: 記載なし
勤務時間: Flex time with no core time
休日・休暇: Five-day workweek : Sat & SunNational HolidaysDate set by the company: Year-end and New Year holidays, etc.Paid leaveSpecial leave: condolence leave: marriage, Spousal childbirth, bereavement leaveRefreshment leaveSummer vacation: 3days
福利厚生: For those who live overseas, we will provide Visa supportrented company housingEach employee pays his/her own rent.Not available during the trial period
社会保険: Unemployment insuranceIndustrial Accident Compensation InsuranceHealth insuranceEmployees' Pension Insurance
掲載日: 2026/10/02 13:17（4日前）
最終確認: 2026/10/05 17:44（1日前）

The Applied AI Engineer, Human Learning role builds a product that uses AI to fundamentally change how people learn and make decisions. What we need is not someone to implement already-defined AI features. Your job is to use learners’ actual behavior, comments, videos, audio, and learning history to answer questions such as the following and turn those answers into a working AI system: - What is happening to this person right now? - What do they understand, and what do they not yet understand? - Why did they make that decision? - What should we teach this person right now? - How should we communicate it so that the way they perceive and judge the situation changes? For example, two learners may both have failed to sauté their onions sufficiently, but for different reasons. One may not understand why moisture needs to be cooked off. Another may be afraid of burning the onions and may have turned the heat down too far. Or the learner may believe they cooked the onions sufficiently even though the video shows otherwise. It is not enough to detect visible failures. We need to infer the understanding and judgment behind them, decide which single issue among several should be taught now, and determine whether that learning carries over into the learner’s next actions. In this role, you will own the full cycle: understanding user problems, designing the AI system and user experience, defining evaluation, implementing it in a real product, and continuously improving it.

moment operates in home kitchens, where the conditions are different every time. Ingredients, cookware, heat levels, quantities, camera angles, sound, and learner skill all vary. From incomplete information—video, audio, and the learner’s own comments—we need to infer what actually happened. Moreover, we need to understand more than the behavior itself. We must also infer states that cannot be observed directly: what the learner understands and why they made a particular decision. Cooking has no single ground truth. Good judgment changes with the ingredients, the goal, and the learner’s level of mastery. Ultimate success is not that the AI returns the correct answer, but that the learner’s way of seeing and judging changes, enabling them to cook better the next time.

In the uncontrolled environment of a home kitchen, infer what actually happened from incomplete information such as video, audio, and the learner’s own comments. Design what should be treated as observed fact and what should be inferred, including cases where the learner’s self-assessment differs from what the video shows.

Infer latent states that cannot be observed directly—what the learner understands, what they do not understand, and why they made a particular decision—from their observable behavior. Continuously model the learner’s current state based on their past learning and behavior.

Elicit the tacit knowledge of world-class chefs and structure it as a Reasoning Framework that AI can apply across different dishes and situations. Then select the single most important thing to teach from among multiple opportunities for improvement, and design how to communicate it so that the learner’s way of seeing and judging changes.

When the right answer varies by dish and learner, define what constitutes a good diagnosis, good coaching, and genuine change in understanding. Evaluate not only the quality of the AI’s response, but also whether the learner’s own judgment and behavior change in their next cooking session.

この求人は、クックパッド株式会社の採用ページの内容をAIが読み取って整理したものです。応募の前に採用ページの原文を確かめてください。データの集め方企業サイト