APEGA EIT
Engineer with hands-on experience building and deploying production ML applications using Python, Azure ML, and neural networks. Demonstrated ability to develop end-to-end AI solutions from model training to web deployment, with focus on model interpretability and real-world business impact. Strong foundation in data science, cloud computing, and full-stack development.
Machine learning system predicting crime severity across 77 Chicago community areas using XGBoost (R² = 23.3%). Engineered end-to-end pipeline from 490K+ crime records (2023-2025) to interactive Power BI dashboard with 2.25M predictions for 2026. Live Dashboard | Source Code
Tools: Python (XGBoost, Pandas, SHAP), Power BI, DAX | 2026
Data Engineering: Built 14-stage ETL pipeline processing 490K crime records. Engineered temporal features (3-hour time blocks, weekend patterns) and severity scoring system (1-10 scale). Cleaned and validated data through multiple quality checks.
Model Development: Trained XGBoost regression model with 5 core features achieving 23.3% R² on zero-heavy dataset (57.6% zero-crime blocks). Used SHAP analysis to quantify feature contributions (Community Area: 64.3%, Time Block: 13.5%). Optimized hyperparameters for production deployment.
Visualization & Deployment: Generated 224K+ daily predictions across all areas and time blocks for 2026. Built interactive Power BI dashboard with temporal slicers, risk categorization, and geographic heatmaps for police resource allocation.
Impact: Enables data-driven deployment decisions by identifying high-risk areas and time periods. Model explanations provide actionable insights: location accounts for 64% of predictions, time patterns 29%. Framework applicable to other cities with crime data.
A Flask-based web app that predicts U.S. health insurance costs using a linear regression model trained with Azure Machine Learning (R² = 0.95). Built with data from Kaggle and deployed on Heroku. Live Demo | Source Code
A Flask-based app that predicts SBA loan outcomes based on user-provided business details. Trained using Azure Machine Learning with a neural network model with a Kaggle dataset. Live Demo | Source Code
B.Sc. Petroleum Systems Engineering
University of Regina | 2023
Email: aprpick@gmail.com
LinkedIn: linkedin.com/in/a-p-0b319520b
GitHub: github.com/aprpick