A significant academic project I led at Loyola University Chicago was Futstat, a full-stack Premier League analytics platform developed as part of COMP 330 – Software Engineering. As Project Lead, I oversaw the entire development lifecycle, coordinating a team of five developers, establishing project architecture, managing version control workflows, and ensuring alignment between frontend, backend, and data science components. The platform was designed to provide live match scores, intelligent predictions, and comprehensive historical team statistics through a modern, interactive dashboard. Built using React.js for the frontend, Node.js with Express for the backend, and PostgreSQL for structured data management, the system integrates real-time football data feeds from external APIs to deliver up-to-date Premier League information spanning the last 25 years.

Beyond project management, my technical contributions centered on backend development, database architecture, machine learning implementation, and the predictions feature. I was responsible for configuring the PostgreSQL database infrastructure, establishing the connection between the Express API and the database layer, and ensuring secure, efficient data retrieval through optimized SQL queries. I architected the database schema to support complex queries across multiple normalized tables containing historical match data, team statistics, player performance metrics, and fixture information, all populated through Python ETL pipelines I developed for automated data ingestion and transformation. A primary focus of my work was designing and implementing the machine learning prediction system from the ground up. I developed supervised learning models using Python’s scikit-learn framework, training on 25 years of historical Premier League data to forecast match outcomes. The feature engineering process involved extracting and constructing relevant predictors including rolling team form metrics (goals scored/conceded over trailing windows), head-to-head historical performance, home/away advantage indicators, rest days between fixtures, and weighted seasonal performance trajectories.

I experimented with multiple classification algorithms including logistic regression for interpretability, random forests for handling non-linear interactions between features, and gradient boosting methods (XGBoost) for maximizing predictive accuracy through ensemble learning. The model training pipeline incorporated cross-validation techniques to prevent overfitting, with temporal splits ensuring training data always preceded test data to avoid data leakage. I implemented hyperparameter tuning through grid search to optimize model performance, evaluating results using metrics such as accuracy, precision, recall, and log loss for probability calibration. The final production model outputs class probabilities for home win, draw, and away win scenarios rather than hard classifications, providing users with nuanced confidence intervals. I serialized the trained models using pickle for efficient loading in the production environment and integrated them into the Express backend through spawned Python child processes, enabling real-time inference via RESTful API endpoints.

On the frontend, I designed and developed the predictions tab to visualize model outputs effectively. The interface displays upcoming fixtures with probability distributions represented through progress bars and percentage metrics, alongside historical accuracy statistics to build user trust in the predictions. I integrated interactive elements allowing users to explore feature importance and understand which factors most significantly influenced each prediction, making the model’s decision-making process transparent and interpretable. The platform architecture emphasized separation of concerns with clearly defined API routes for match data retrieval, prediction serving, and team analytics. I implemented environment-based configuration management for secure credential storage, CORS middleware for cross-origin communication, and comprehensive error handling to ensure system reliability. The backend served as an orchestration layer, coordinating between the PostgreSQL database for historical queries, external football APIs for live data, and the machine learning inference engine for predictions.

Throughout development, I established coding standards, conducted code reviews, managed pull requests, and maintained comprehensive technical documentation including system context diagrams, container architecture diagrams, and dataflow visualizations. I ensured the codebase followed best practices such as parameterized SQL queries to prevent injection attacks, modular component design for maintainability, and version-controlled environment configurations for reproducible deployments. This project demonstrated comprehensive full-stack and machine learning engineering capabilities, encompassing data pipeline development, feature engineering, model training and evaluation, API integration, database optimization, and technical leadership in a collaborative software engineering environment.


This is the link and the github link to the project:
https://premier-league-analytics-comp330.vercel.app/
https://github.com/osanchezhuezca/Premier-League-Analytics-Platform