Ervin Samuel, 陳才兵

    


[email protected]
0961 100 735Hsinchu, Taiwan

Work Experience

June 2021 — August 2021

Software Engineering Internship at HTC

  • Responsible for writing automated tests using Robot Framework, a test automation framework based on Selenium.
  • Worked on a UI test for the VIVE UK home page that verifies texts are correct, images can load properly, links and buttons lead to a working page, and forms are able to submit.
  • Worked on an API test for the VIVE UK Accessories page that verifies the product prices displayed are the correct prices received from the API.

Education


September 2018 — Present, expected to graduate in June 2022.

BS, Computer Science, National Tsing Hua University

Current GPA - 3.97
Relevant Skills and Courses:
        — Python, C++, C, SQL                     Data Structures and Algorithms
         Keras, Pytorch, Scikit-learn         Git Version Control
        Data Science                                 Database Systems

Projects

3D Tic-Tac-Toe with Monte Carlo Tree Search


Applied the Monte Carlo Tree Search with Upper Confidence Bound applied for Trees on a 4 by 4 by 4 Tic-Tac-Toe state space to create an agent that calculates the next optimal move which resulted in a more than 100%-win rate when faced against a Random Agent.

Image Classifier with ResNet

Developed an image classifier in python using the fast.ai library with training data taken from Bing image search by fine-tuning the ResNet model, which resulted in a prediction accuracy higher than 90%.

Tic-Tac-Toe agent with Reinforcement Learning

Designed a Tic-Tac-Toe agent by learning state values from over 50,000 iterations of playing against itself and applying the Monte Carlo algorithm, which resulted in an unbeatable agent.

COVID-19 30-day Mortality Prediction from CXR

Built a CNN model by using Keras in the Tensorflow library with three hidden layers, which predicts a 30-day mortality from CXR which had a 0.75 F1 score.

Pathfinding Visualizer

Designed and built a program with python to visualize pathfinding in a randomly generated maze using Kruskal's Algorithm, Bellman-Ford and A* Search algorithms.