Harshatha Prasanna

hprasanna@ucsd.edu | (925) 202-5868 | LinkedIn

Summary

UC San Diego Data Science undergraduate with hands-on research experience in responsible data management, data cleaning, and statistical analysis. Passionate about transparency and fairness in algorithmic decision-making, with a data-centric approach to building tools that empower human understanding of complex datasets.

Education

University of California, San Diego

Bachelor of Science in Data Science; Minor in Cognitive Science | May 2028

GPA: 3.83
Relevant Coursework: Data Structures and Algorithms, Statistical Analysis, Research Methods, Programming for Data Science, Principles of Data Science

Skills

Programming: Python, Java, SQL, SQLite, R
Data Analysis: Exploratory Data Analysis (EDA), Statistical Modeling, Data Visualization, Hypothesis Testing
Machine Learning: Supervised & Unsupervised Learning, Feature Engineering, Model Training & Evaluation
Libraries & Frameworks: Scikit-learn, Pandas, NumPy, Matplotlib, Seaborn, Pytorch, Keras
Data Tools: Jupyter Notebook, Git, Tableau

Experience

Responsible Data Management Research

Independent Researcher | San Diego, California | November 2025 – January 2026

Analyzed American Community Survey data (1.6M+ observations) using Python to identify patterns across demographic and occupational variables, uncovering a 29% income disparity across occupation codes.
Identified a critical data infrastructure gap where existing datasets document outcomes but omit mechanism variables needed to explain them, highlighting a fundamental limitation in data-driven decision-making pipelines.
Developed a fully reproducible data pipeline with statistical analysis and visualizations, published on GitHub to promote transparency and accessibility of findings.

UC San Diego Data Science Society (DS3)

Data Analyst | San Diego, California | January 2025 – May 2025

Collected, cleaned, and preprocessed CDC NHANES dataset (12,000+ participants, 34 features) to support reliable model development and ensure data quality throughout the pipeline.
Engineered and optimized an XGBoost classifier using hyperparameter tuning and Repeated Stratified K-Fold cross-validation, achieving 96% cross-validated accuracy and delivering a 26% increase in recall and 61% increase in precision compared to baseline.
Evaluated model performance metrics to support transparent and trustworthy algorithmic decision-making.

UC San Diego CSE Society

Technical Writer | San Diego, California | September 2025 - Present

Conduct interviews with researchers, faculty, and student teams to gather technical information and accurately represent their work across multiple publication formats.
Translate research into accessible content for newsletters and campus publications across diverse audiences.
Participate in brainstorming sessions to develop innovative approaches to present research to broad audiences.