Gino Angelici

Irvine, CA

Built an end-to-end LLM pipeline with prompt engineering and fine-tuned classification models to generate AI digital twins simulating survey responses across political subgroups, using iterative re-prompting to correct bias; validated with Jensen-Shannon divergence and chi-square tests of homogeneity against real respondents.

Python · LLM Prompt Engineering · Fine-Tuned Classification Models · Statistical Validation

Autonomous vehicle development

Designed CNN-based perception and control models for autonomous vehicle navigation on low-power embedded hardware, integrating stereo and RGB-D cameras; trained across 3 prototype configurations on UCSD's Supercomputer Center and built MLOps workflows in Weights & Biases; benchmarked against LLMs/LVMs and led the team to 1st place at Purdue's national AV competition.

PyTorch · Computer Vision · CNNs · MLOps (Weights & Biases) · Embedded Systems

Built Python ETL pipelines to analyze 70+ years of climate records across 60 airports, applying PCA, polynomial, and ridge regression to identify statistically significant correlations; contributed to a peer-reviewed publication linking urbanization to local temperature change.

Python · ETL Pipelines · PCA · Polynomial & Ridge Regression · Climate Data Analysis

Trained and evaluated XGBoost and regularized regression models on 50,000+ HHS records to surface predictors of substance-abuse risk, presenting findings to senior leadership via Plotly dashboards.

XGBoost · Regularized Regression · Feature Importance · Plotly