~/allan/home/about.ap
~/allan/home/links.ap
~/allan/home/work.ap
main
../about.ap
© 2025 Allan Paiz. All rights reserved.
main
../about.ap
© 2025 Allan Paiz. All rights reserved.
I'm currently a student at theUniversity of South Carolina. Computer Science major, Mathematics minor. Expected graduation December of 2026. I enjoy all sports with (and without) the use of analytics. soccer>(football>=baseball)>(hockey>=basketball)I enjoy playing tournament poker and competitive video games, though I don't get the chance to play as often as I'd like. United States Marine Corps Veteran. When I'm not on a computer... I'm enjoying life with my wife, twin babies, and 2 doggos.
Defensive Stopping Power : NFL Big Data Bowl 2024
- New defensive metrics using NFL tracking data. Wash Your Rice
- Horror themed puzzle game. Dynamical Systems and AI Research Group
- Research in robotics, AI, and dynamical systems. TuneUP by JETA
- Music application for my software engineering course. Predicting Construction Permit Delays
- Math Foundations of ML/DS course project. Green Dog : NFL Big Data Bowl 2025
- A tool that predicts the assignment of an NFL receiver. // This portfolio is meant to resemble a neovim text editor. // It looks great on my PC, but may not on yours... // if that's the case, we're UNDER CONSTRUCTION!
at] gmail [dot] comGithub: https://github.com/allanpaizResume: /resume.pdfUpdated: 8/19/2026
2024 NFL Big Data Bowl competition's tackling theme left me struggling to find a direction for my project.
Several of my ideas were limited by the provided datasets.
first_contact flag in the data was misleading. (See GIF)identify all contact, or potential contact, between ball carriers and defenders.
Links: GitHub Repository Kaggle Submission Finalists: NFL Announcement Kaggle Announcement Summary: Developed a neural network to detect all defender-ball carrier contact in NFL tracking data, addressing limitations in traditional defensive metrics. Technical Skills: Python, Pandas, NumPy, Keras, Matplotlib Personal Skills: data processing, problem solving, data analysis, machine learning, feature engineering, visualization, reporting Features: feed-forward neural network, spatial and velocity data modeling, detailed visualization and analytics Applications: NFL and NCAA player scouting and evaluation, game planning and analysis, in game decision making, Fantasy football and sports betting analytics
Wash Your Rice is a five-person capstone built in Unreal Engine 5.6.1 and C++. The technical work is a collection of connected systems: an in-world computer, puzzle progression, survival state, night orchestration, UI, and verification.
I built core parts of the in-world computer interface and its puzzle-progression flow in UE5 C++, then added automated coverage for interface state transitions and puzzle validation.
Links: Capstone Site GitHub Repository Final Demo Proof of Concept Summary: Built core C++/UMG computer-interface and puzzle-progression systems for a five-person Unreal Engine 5.6.1 capstone, with automation coverage for UI state and puzzle behavior. Technical Skills: Unreal Engine 5.6.1, C++, Rider, UMG, Enhanced Input, Slate, GameInstanceSubsystem, Unreal Automation Testing, Git LFS Personal Skills: systems design, debugging, integration, test-driven iteration, team collaboration Engineering Focus: persistent game state, computer UI and progression, inventory and sanity components, timer-driven events, save/load, UI flow, regression coverage
Dynamical Systems & AI Research Group under Dr. Vignesh Narayanan.
The work connected robotics, control, optimization, and AI through experiments with autonomous navigation and perception systems.
Summary: Robotics research spanning TurtleBot4 experiments, safe navigation, optimization and control, multi-agent simulation, and perception-system design. Technical Skills: ROS 2, Gazebo, TurtleBot4, OAK-D, 2D LiDAR, YOLOv8, Python, constrained optimization, control theory Personal Skills: experimental design, research communication, systems thinking, technical writing Focus Areas: obstacle avoidance, event-triggered optimal control, multi-agent path finding, sensor fusion, safety-aware robotics
TuneUp is a prototype music app developed in Java for my Software Engineering course.
Teamwork - Worked in a team of four to complete a project from start to finish.Planning - Engaged with a client to define project requirements and objectives.System Design - Created documentation and UML diagrams to guide development.Development & Testing - Built and tested a Java console app, using GitHub for version control and collaboration. Front End - Designed and developed a GUI with JavaFX, applying UI/UX principles.
Links: GitHub Repository Summary: Developed a music application for teacher-student interaction with composition and free play modes, lessons, and song libraries. Technical Skills: Java, JUnit, JavaFX, git, GitHub, Object-Oriented Programming, documentation Personal Skills: team collaboration, communication, project management, presentation Features: software development life cycle, system design, software engineering, unit testing, front end development
can permit timing help a project manager estimate schedule risk while preparing a bid?
Delays in permit issuance can change a construction schedule before work begins. I framed that uncertainty as a classification problem: flag issued residential permits that are likely to take longer than normal.
Links: City of Atlanta Building Permits Data Summary: A 7,504-record permit-issuance-delay classifier that compared a location-only RBF SVM, structured neural network, and text-enhanced neural network. The description model achieved 69.06% test accuracy and 77.44% high-risk sensitivity. Technical Skills: Python, scikit-learn, RBF kernel SVM, neural networks, TF-IDF, one-hot encoding, random oversampling Personal Skills: problem framing, model comparison, analytical communication, responsible interpretation Methods: temporal train/test split, public-data preprocessing, structured permit modeling, description-text features, classification metrics
Green Dog was my 2025 NFL Big Data Bowl project. It grew from a defensive question: can a pre-snap tool help predict whether an eligible receiver will run a route or stay in to block?
That decision can change a defender's assignment and help reveal an offense's tendencies before the ball is snapped.
Links: GitHub Repository Kaggle Submission Summary: A pre-snap model that estimates whether an eligible receiver will run a route or stay in to block, giving football decision-makers another way to inspect tendencies. Technical Skills: Python, TensorFlow, Keras, pandas, NumPy, scikit-learn, Matplotlib, Seaborn Personal Skills: feature engineering, model evaluation, sports analytics, technical storytelling Features: pre-snap NFL tracking data, play/player/team feature branches, route probability, week-by-week evaluation