Hello, I'm
Hiroto Ishii
I'm a fourth-year Management and Information student building practical software and machine learning systems. My focus is turning vague ideas into working prototypes: data-driven applications, neural network experiments, and tools that make technical concepts easier to use.
/ Career
A brief overview of my career
BSc Student, Management and Information
University of Shizuoka
Studying information systems, programming, data analysis, and management. Currently working on a graduation research project around neural network training dynamics and parameter sharing.
Neural Network Research Project
Graduation Research
Implementing and evaluating neural network structures that begin with shared parameters and gradually differentiate during training. Building experiments in C++/Python and comparing learning behavior against independent-neuron baselines.
Organizer / Study Session Lead
Engineering Circle
Planning beginner-friendly engineering activities, including programming contests and machine learning workshops. Designing materials that help members move from theory to hands-on implementation.
BSc Student, Management and Information
University of Shizuoka
Studying information systems, programming, data analysis, and management. Currently working on a graduation research project around neural network training dynamics and parameter sharing.
Neural Network Research Project
Graduation Research
Implementing and evaluating neural network structures that begin with shared parameters and gradually differentiate during training. Building experiments in C++/Python and comparing learning behavior against independent-neuron baselines.
Organizer / Study Session Lead
Engineering Circle
Planning beginner-friendly engineering activities, including programming contests and machine learning workshops. Designing materials that help members move from theory to hands-on implementation.
/ Featured Projects
A selection of professional work, published apps, and experimental prototypes.
A research prototype for testing whether a neural network can start with shared hidden units and gradually split them based on training dynamics. Built with C++ and Python to analyze convergence behavior and compare against independent-neuron baselines.
A product idea for highlighting risky clauses in terms of service documents. The design emphasizes source-specific highlighting, clearer scoring, and risk evaluation based on maximum severity, frequency, and average risk.
/ Tech Stack
Breakdown of my technical skills, categorized by domain and proficiency
Machine Learning / Data
Programming
Frontend / App
Tools / Infrastructure
/ Contact Me
Feel free to reach out. I'm always open to discussing new projects and opportunities.