Takumi Horiba

Undergraduate Student in Computer Science & Statistics at UBC

About

Hello, I'm a 4th year student at the University of British Columbia studying computer science and statistics. I have experience in reporting using SQL and SAP BusinessObjects, data analysis, and machine learning model development and evaluation with PyTorch. Please check LinkedIn and/or CV for more up-to-date info.

Skills

Programming: Python, R, Java, C, C++, TypeScript, SQL
Tools: Git, Linux command line, Node.js
Data Analysis: SQL, pandas, numpy, sci-kit learn, Jupyter
Statistics: statistical inference, linear models
Other: AWS EC2

Experience

Junior Data Scientist

TRIUMF, Vancouver, BC. May - Dec 2024

- Developed and optimized deep learning models: Trained ResNet models for neutrino event classification and regression, improving model robustness against systematic uncertainties using PyTorch.
- Enhanced statistical data analysis tools: Developed reusable analysis code to improve speed and quality, reducing manual work and minimizing errors while evaluating the performance of over 50 tests simultaneously
- Implemented custom dataset class in PyTorch: Enabled the artificial creation of noise in data for training and testing robust ML model to uncertainty from experimental data, reducing statistical bias by nearly 20%.
- Gained proficiency in Linux environments, CUDA, HDF, and computer clusters to handle big data effectively
- Part-time work from Sept. - Dec. 2024
GitHub

Co-op Student, Business Intelligence Solutions

Provincial Health Services Authority, Vancouver, BC. May - Dec 2024

- Developed and implemented a KPI dashboard to enhance the team's understanding and monitoring of service performance metrics, aiding in operational decision-making processes.
- Created an Immunization Dashboard which streamlined the retrieval process for vaccine-related data requests, enhancing response times and supporting public health initiatives.
- Improved public speaking through presentations and demo for multidisciplinary audiences ranging from small teams to large groups of 100 attendees.

Mathematics Teaching Assistant

University of British Columbia, Vancouver, BC. Sep 2022 - Apr 2023

- Guided and helped lead workshops with up to 30 students to deepen their mathematical skill in collaboration with a graduate math TA.
- Improved clear oral communication skill to support student success in the process of identifying student concerns by hearing them and observing their work.

Projects

Past Course Data Dashboard

  • Developed web-based dashboard using React and Typescript
  • Implemented APIs and data wrangling workflow in TypeScript that converts raw data (in zip files of HTML and JSON) into structured format in JSON that can be more easily used.
  • Worked in a team of two with Git and followed agile methodologies.

Integrated Bus Departure Board

  • Develop a website that shows all real time bus departures from UBC exchange using node.js
  • Process data from API and sort it in chronological order with details such as delay and bay info
  • Employ intuitive coloring and design for critical info such as route number and delay info by customizing CSS with Bootstrap framework.
    Technologies: Node.js, JavaScript, JSON, HTML/CSS

Statistical Inference Project on Airbnb Accommodation Price

  • Performed ANOVA and two-sample t-test to verify hypotheses on groups of population data using R
  • Compared the results between bootstrap based and normality based approaches
    Technologies: R, Jupyter, Git, R libraries (infer, ggplot2)
    • GitHub

Exercise Log

  • Java application that can log user's exercise history and give recommendations based on their history.
  • Supports data perseverance, console UI, and GUI through JSON and Java Swing library
  • User's actions are recorded and displayed when they close the application
  • Test-driven development with JUnit.
    • GitHub

Heart Disease Prediction

  • Predict presence of heart disease by basic characteristics of a potential patient such as sex, age and cholesterol level.
  • Use R and its packages to get, clean, analyze and visualize data in collaboration with 3 teammates.
  • Create a reproducible electronic report that shows our model and results.
    • GitHub

Tic-tac-toe Player

  • Web application to play tic-tac-toe against an algorithm implemented in Python.
  • Minimax algorithm is implemented in PyScript and and deployed using GitHub pages.
    • GitHub

Stock Trading Simulator

  • Web application to to simulate buying and selling of stock by Python, flask and JavaScript
  • Data is queried through API(IEX API) each time user wants.
  • SQLite database stores user information such as balance, username, and password.

Education

Courses at UBC

Relevant courses

Computer Science Statistics and Data Science Mathematics
  • MATH 200 Multivariable Calculus
  • MATH 221 Matrix Algebra
  • MATH 303 Stochastic Processes
  • MATH 307 Applied Linear Algebra
Economics
  • ECON 101 Principles of Microeconomics
  • ECON 102 Principles of Macroeconomics

Deep Learning Specialization

I learned various models in deep learning such as CNN, RNN and transformers throughout 5 courses. I built several small applications such as music generation with Python and TensorFlow. This specialization to made me want to apply deep learning techniques for real world situations, and learn more about this growing field of machine learning.
Certificate on Coursera