Tomoya Matsubara
Second-Year Ph.D. Student at Keio University
Majoring in computer science, I have been developing desktop and mobile applications. My current research is about efficiently updating high-quality 3D Gaussian Splatting scenes based on change detection.
Desktop Applications
Mobile Applications
Web Applications
Publications
- Change Detection for Constantly Maintaining Up-to-date Metaverse Maps
- Tomoya Matsubara, Maki Sugimoto, Hideo Saito
- 2024 IEEE Conference on Virtual Reality and 3D User Interfaces Abstracts and Workshops (VRW). IEEE, 2024.
- Point Cloud Change Detection In Indoor Environments
- Tomoya Matsubara, Hideo Saito
- CEUR Workshop Proceedings. Vol. 3467. CEUR-WS, 2023.
- Shared Transformer Encoder With Mask-Based 3D Model Estimation For Container Mass Estimation
- Tomoya Matsubara, Seitaro Otsuki, Yuiga Wada, Haruka Matsuo, Takumi Komatsu, Yui Iioka, Komei Sugiura, Hideo Saito
- ICASSP 2022-2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2022.
Work Experience
Machine Learning Engineer Internship in Product Development Department
Howtelevision
11/2022 - Present
Tokyo, Japan
- Updated a machine learning model to classify the category of questions posted on the company's Q&A platform and improved the accuracy by 4%.
- Developed a machine learning model for assigning highly specialized questions to experts, reducing manual work of assigning questions by 50%.
- Introduced MLOps pipelines to automate the repetitive process of training and deploying machine learning models with TensorFlow Extended (TFX) in Google Cloud Platform (GCP).
- Developed an API server with FastAPI to serve all of the machine learning models used in the company's Q&A platform.
Backend Engineer Internship in Common System Development Department
Dwango
09/2022 - 03/2023
Tokyo, Japan
- Developed a web application that enables non-technical staff to create and manage internal APIs. Used Go for the backend and Next.js for the frontend.
- Developed batch jobs to gather user activity data from the company's services, utilizing SnowSQL for subsequent analysis.
- Created APIs for mobile application notification services using Ruby.
- Conducted an inspection of the middleware utilized by the company's services for data streaming, leading the migration from RabbitMQ to AWS services to reduce operational costs.
Cloud Support Engineer Internship
Amazon Web Services
01/2023 - 01/2023
Tokyo, Japan
- Handled customer issues related to AWS services such as EC2, ALB, and S3.
Data Scientist Internship
GMO Internet Group
10/2022 - 10/2022
Tokyo, Japan
- Developed machine learning models to predict the click-through rate (CTR) of ads, achieving an approximately 80% AUC score using imbalanced datasets containing millions of rows.
Automotive System Engineer Internship in Cross-Domain Computing Solutions Division (XC)
Bosch
08/2022 - 09/2022
Ho Chi Minh City, Vietnam
- Refactored and cleaned up the codebase of an automotive system written in C to improve readability and maintainability and to reduce security vulnerabilities.
- Developed a self-driving car system by tuning the parameters in the simulation environment.
Data Scientist Internship in Digital & Integration Department
Schlumberger
07/2022 - 09/2022
Tokyo, Japan
- Analyzed data from technical support requests for the company's products over the past 4 years to identify the most common issues and their root causes, providing valuable insights into the company's product development and customer support.
- Analyzed software license contracts of the company's products over the past 4 years to get a deeper understanding of the company's customers' needs and to predict optimal times for advertising the company's products to new customers.
Backend Engineer Internship
Howtelevision
08/2022 - 08/2022
Tokyo, Japan
- Collaborated with the job search platform team to design and develop APIs in Go for creating user-friendly pages that enable users to manage their information regarding past and current job applications.
- Replaced existing backend PHP code, which had accrued technical debt, with a new Go backend as part of a project aimed at enhancing service interactivity with React/Go.