Principal Applied AI Engineer（moment by cookpad）

Job Description

About Cookpad

Cookpad develops the world’s most user-friendly products for people who cook.
The use of AI and robotics broadly falls into two approaches: replacing human actions, or extending human capability. Our focus is on using these technologies to bring out people’s strengths.
Cooking affects our health, cognition, society, culture, and the environment. Our challenge is to make it possible—through our products—for anyone, anywhere, to cook at an unprecedentedly high standard.

The essence of this position

moment

helps people learn to cook in an innovative way with a personal coaching service. This service is completely based in AI using multimodal (text, vision, audio). The challenges at moment are not about incremental or partial improvements. What is required is the ability to identify the gap between the ideal learning experience and what AI can realistically deliver today, and then to narrow that gap by isolating and fully solving the single point where the greatest impact can be made right now.

Examples of Key Challenges

Video Analysis Domain: An elite coach can instantly discern from video the heat level, early signs of failure, and the appropriate corrective actions in cooking. The goal is to reproduce and extend this judgment using AI.

This is not a job focused on improving accuracy metrics, but on implementing the judgment of top-tier experts themselves.

• Solving fundamental challenges in understanding cooking videos
• Designing multimodal (video, audio, text) decision-making models
• Designing coaching logic that takes conversational memory and learning state into account
• Designing and implementing task / research agents

Responsibilities

• Work backward from the ideal learning experience to independently define the most critical problem
• Translate the problem into technical requirements and lead the design, implementation, and validation end to end
• Be able to explain how the solved problem impacted learning speed, retention, reproducibility, and cost

Requirements (Must-Haves )

• Approximately 5+ years of software development experience
• Strong professional proficiency in Python
• Hands-on experience developing and integrating applications using LLMs
• A proven track record of defining problems from the gap between the ideal and reality, discarding many possible improvement ideas, and focusing on the single most impactful point to deliver results.
• Experience rapidly iterating through hypothesis, design, implementation, and validation for problems with no single correct answer
• Practical knowledge of optimizing generative model outputs
• Understanding of modern development practices (TDD, version control, CI/CD, containerization, etc.)
• Ability to translate user context and emotions into technical specifications

Preferred Qualifications

• Excellent English communication skills within cross-functional teams
• Experience deriving insights from user behavior or conversational data to inform AI design requirements
• Background in computer vision or traditional machine learning (especially with video data processing)
• Experience developing hybrid solutions combining modern generative AI and traditional methods
• Sensitivity to human–AI interaction nuances and user experience optimization
• Practical understanding of AI model trade-offs, performance, and constraints
• Experience with vector embeddings and vector databases (e.g., Pinecone)
• Experience implementing RAG architectures for LLM-based applications
• Experience fine-tuning and customizing large language models using frameworks such as LangChain, LlamaIndex, Hugging Face, AWS Bedrock, or Google Vertex AI
• Experience developing in cloud environments (AWS, GCP, or Azure)
• Knowledge of MLOps / LLMOps (model deployment, serving, monitoring, etc.)
• Foundational understanding of data engineering, such as building data pipelines for AI applications