Xinming Mia Dai

• Assistant to Professor Marc Scott improving frequency plots by employing Hamming distance to cluster analogous sequences in sequence analysis

• Developed four interval estimation methods for zero-inflated gamma distributions based on fiducial inference, Box-Cox transformation, parametric bootstrap, and the method of variance of estimates recovery (MOVER); designed and conducted computer experiment to compare the performance of the four estimation methods (code can be found here)
• Designed confidence intervals constructed using a pivot quantity for semiparametric single-index models; carried out simulations for these new methods in R
• Established a novel addable ordinary differential equation model with the group MCP penalty for reconstruction of gene regulatory networks with data from DREAM Systems Biology Challenges

• 2024 Fall UN2103 Applied Linear Regression Analysis
• 2025 Spring UN2104 Applied Categirical Data Analysis

• Constructed a camera with four lenses, supported by a Raspberry Pi chip. The camera is capable of capturing four pictures simultaneously and processing them into a GIF-format wiggle image
• Utilized Haar Cascade classifiers in OpenCV to detect faces in images, relocated pictures around detected faces to generate wiggle images using the Pillow package in Python
• Built a graphical user interface (GUI) using Tkinter in Python, providing an easier and more user-intuitive image processing experience

• Course assistant for Multilevel Modeling for 60 Master’s and PhD students; explained hierarchical linear model concepts and LMER syntax in mini lab sessions during office hours
• Lab instructor for Statistical Computing class; prepared lab activities focusing on practicing statistical analysis, graphic representations, and statistical simulation using R

Coluxa Inc. is developing an innovative microscope system to support life science research and disease diagnosis.

• Developed innovative deep learning algorithms to analyze fluorescent markers in human cells, revamping an otherwise labor-intensive manual process and develop a new AI-assisted microscope
• Automated bioimage processing and analysis by tuning algorithm parameters and implemented using TensorFlow
• Processed image data with OpenCV and sped up model training by 40% by optimizing data delivery using CUDA
• Built visualizations with Seaborn to evaluate and enhance algorithm performance and analyze scanned objects
• Slides

• Optimized and automated data collection processes by 70% (in time) by building web scrapers with Python urllib and parsed real estate market data with Beautiful Soup
• Conducted market research and collected 500+ pieces of real estate data in collaboration with stakeholders
• Cleaned and synthesized messy data using Pandas; developed internal PostgreSQL databases by collaborating with the data and consulting departments

• Coursework: Algorithms, Machine Learning, Causal Inference, Multi-Level Modeling, Experimental Design
• Leadership Activity: Society for Statistics Club Board Member (American Statistical Association)

• Math Coursework: Real Analysis, Algebra, Mathematical Analysis, Mathematical Optimization, Differetial Equation
• Statistics Coursework: Probability, Time Series, Stochastic Process
• Computer Science Coursework: Algorithms, C++, Matlab, Statistical Computing
• Finance Coursework: Microeconomics, Macroeconomics, Life Insurance Actuarial Science