Introduction
During my undergraduate studies, I had the opportunity to work as an AI engineer in two different environments: an educational company and an AI startup. Through these experiences, I worked on real-world problems involving machine learning and optimization, and gained both technical and practical insights.
In this article, I will share what I worked on, the challenges I faced, and what I learned from these internships.
Internship 1: AI Engineer at Nagase Brothers Inc.
Project Overview
At Nagase Brothers Inc., I worked as an AI engineer on a project to predict the average scores of the Japanese National Center Test (Common Test).
Predicting the average score before the official announcement is important for two main reasons:
- Immediately after the exam, students estimate their scores and compare them with predicted averages to decide which universities to apply to.
- Accurate predictions demonstrate the credibility and analytical strength of the preparatory school.
Although the preparatory school operated by Nagase (Toshin) is a major player, it has fewer students compared to other large institutions, meaning that the available dataset is relatively limited. Therefore, our goal was to achieve high prediction accuracy using limited data through a combination of domain-specific logic and machine learning techniques.
Key Challenge: Curriculum Change in 2025
My biggest contribution was handling the major curriculum reform introduced in 2025.
The structure of the exam changed significantly, including:
- New subjects being introduced
- Changes in subject composition
- Incompatibility with historical data
This required redesigning the prediction system, especially:
- Mapping old subjects to new ones
- Reconstructing features based on limited historical data
- Ensuring model robustness despite distribution shifts
Result
By conducting thorough exploratory data analysis and iteratively refining our approach, we successfully adapted the system to the new curriculum.
As a result:
- Our prediction ranked #1 in accuracy among major preparatory schools
This experience taught me how to handle real-world data limitations and adapt machine learning systems to structural changes.
Internship 2: AI Engineer at Emuni Inc.
Background
Motivated to further develop my skills, I joined Emuni Inc., a startup originating from the Matsuo Lab, as a contracted AI engineer.
Project Overview
The project focused on production planning optimization for a chemical manufacturing company.
Unlike my previous experience, this project involved:
- Developing a system for external clients
- Collaborating with engineers across multiple domains (frontend, backend, and ML)
- Building a production-level application
My Role and Contributions
I worked primarily as an ML engineer, where I was responsible for:
- Designing and implementing machine learning components for optimization
- Assisting in requirement definition with the client
- Integrating ML models into the overall system
The ML component played a critical role in improving the efficiency of production planning.
Technical Experience
Through this project, I gained hands-on experience with:
- Web application development
- Cloud-based debugging using Azure
- Handling data with Azure Blob Storage
- Cross-functional team development
This was my first experience working in a highly collaborative engineering environment, which significantly broadened my perspective.
Key Takeaways
Through these two internships, I developed both technical and professional skills:
Technical Skills
- Machine learning under limited data conditions
- Handling distribution shifts and system redesign
- Production-level system development
- Cloud technologies (Azure)
Professional Skills
- Communicating with stakeholders and clients
- Translating business requirements into technical solutions
- Working in cross-functional teams
Future Goals
These experiences strengthened my interest in combining machine learning and optimization to solve real-world problems.
In the future, I aim to:
- Work on large-scale optimization and scheduling problems
- Develop AI systems that create real business impact
- Continue bridging the gap between theory and real-world applications
Conclusion
My internship experiences allowed me to go beyond academic learning and tackle complex, real-world challenges. From handling limited data in education systems to optimizing industrial processes, I gained valuable insights into how AI can be applied in practice.
These experiences have strongly shaped my career goal of becoming an AI engineer who can deliver impactful solutions.