Emily Oh
M.S. Business Analytics @ UCLA Anderson (‘25)
B.A. Economics @ UC San Diego (‘21)
3+ yrs in FinTech, IT, and software, entrepreneurship
Currently diving into AI/ML projects
Born in Germany, raised in Korea - I love exploring innovative technologies, outdoors, cooking, and hanging out with my dogs!
Favorite Quote : “Success is not final, failture is not fata: It tis the courage to continue that counts.” - Winston Churchill
Credit Card Fraud Detection (Data Balancing)
-
Gained experience in GANs to tackle data imbalance in fraud detection by generating synthetic transactions. Learned to build and train models improving the classifier’s accuracy. Strengthened my skills in deep learning and data augmentation techniques.
TensorFlow/Keras: For building and training the GAN model.
Pandas, NumPy: For data manipulation and preprocessing.
Scikit-learn: Utilized PCA for dimensionality reduction and StandardScaler for normalization.
Seaborn, Matplotlib, Plotly: For data visualization (scatterplots and histograms).
Applied PCA for visualizing real and synthetic data in two dimensions.
Implemented GANs for generating synthetic data and resolving class imbalance in the dataset.
Personal Projects
Tesla Stock Price Prediction (FB Prophet)
-
Built a predictive model using Facebook Prophet in Python to forecast Tesla stock prices over a 30-day period. Leveraged time-series analysis to improve the model's accuracy, providing actionable insights for financial analysis and investment decisions.
Facebook Prophet: For time-series forecasting.
Plotly Express: For interactive data visualization.
Google Finance (in Google Sheets): For forecast evaluation and stock analysis.
Yahoo Finance: For retrieving Tesla stock price data.
Time-Series Analysis: Improved the model's prediction accuracy by analyzing Tesla stock price trends.
Forecasting & Evaluation: Predicted future stock prices and evaluated the model’s performance using financial data.
Robot Localization (Particle Filters)
-
Developed a particle filter algorithm in Python and NumPy to solve a robotics localization problem. Simulated real-world conditions with noisy sensor data and imprecise movement, resulting in a highly accurate localization model in an unknown environment.
Python, NumPy: For coding the particle filter and handling data.
OpenCV: For map visualization and robot/particle display.
Keyboard Inputs: To simulate robot movements and control its actions.
Particle Filter Algorithm: Initialized and moved particles based on user inputs, simulated robot movement and sensor readings.
Sensor Simulation & Weight Calculation: Simulated noisy sensors and assigned weights to particles to reflect their accuracy.
Resampling: Improved localization accuracy by resampling particles based on their weights.
Noise Addition: Added random noise to simulate real-world imperfections in sensor and movement data.
Robot Localization Display: Visualized the robot, particles, and estimated positions on the map.
Entrepreneurship
In the summer of 2019, I came up with the idea to build an app where people could buy, sell, and trade crops they grew themselves. I led the development of the app using React Native alongside seven talented UCSD students, allowing users to connect and exchange their homegrown produce.
To bring the project to life, we secured $7,000 in funding from the Blackstone LaunchPad and startup conventions. Along the way, we partnered with 20+ local farmers and growers, expanding the app’s reach and improving its operations.