I'm a driven machine learning and data science student at UC Santa Barbara, passionate about building impactful tools that sit at the intersection of AI and real-world decision-making. My experiences span computational biology, bioinformatics, and applied statistical modeling, with hands-on work in both academic research and large-scale ML projects. I'm especially interested in healthcare analytics, causal inference, and end-to-end ML pipelines. Let's build something meaningful.
Analyzed over 10,000 genomic sequences using Python-based DANPOS tools to investigate nucleosome dynamics and protein-DNA occupancy. Developed EDA pipelines and visualizations to enhance genomic insight communication, improving pipeline efficiency by 8%.
Built interactive R dashboards for genetic simulation data. Designed ETL pipelines using Python and R to automate preprocessing, reducing data wrangling time by 10% and accelerating model iterations for hypothesis testing.
Coursework includes Statistical Machine Learning (Grad), Econometrics, Design of Experiments, Regression Analysis, and Data Science Principles. Member of Honors Program and awarded 1st place in Houston Hackathon.
Semester abroad focused on international economic models, quantitative methods, and European data infrastructure. Strengthened adaptability and global collaboration skills.
Built a supervised ML pipeline on 2.5M hospital records to predict inpatient charges. Combined preprocessing, feature engineering, XGBoost regression, and SHAP explainability. Deployed via Streamlit and optimized with Optuna.
View ProjectDesigned a machine learning system to predict clinical trial success using structured trial metadata and biomedical embeddings. Leveraged transformer-based NLP to extract insights from drug indications, trial descriptions, and biomedical literature. Aimed at reducing R&D risk for biotech pipelines.
View ProjectBuilt a deep learning pipeline using CNNs and pretrained vision transformers to identify infrastructure damage in satellite imagery post-disaster. Automated geotagged severity mapping for emergency teams. Trained on open datasets like xView2 and validated on real flood/fire zones.
View Project