Projects

StudyBuddy

TypeScript(React), JavaScript(React), SQL

• Built and deployed a full-stack mobile app that helps college students find suitable study partners.
• Configured 25+ REST API endpoints and 7 PostgreSQL tables supporting user information and social features.
• Established secure user authentication with PostgreSQL credential verification and client-side token storage.
• Deployed backend on Railway and frontend on Expo. Collaborated in a 3-person Agile team using GitHub.

View on Github

Shattered Continuum (Video Game) (IN BETA PRODUCTION)

JavaScript (Kaboom.js)

Picture from Game:

Play Right Now!

• Created a browser-based 2D-platformer with a procedural chunk-based level system.
• Engineered a system that destroys active physics objects on traversal for smoother gameplay.
• Architected 10 levels alongside random and max-difficulty modes, each with 4 independent feature rolls.

View on Github

Running Workout Generator

C++

Example Input:

Example Output with flags -b -d -p:

• Constructed a C++ program that creates running workouts based on training information.
• Utilized enumerated classes and bash command line options to easily and quickly print out workout options.
• Optimizes the ideal workout based on 6 different inputs (Surface, Distance Goal, Training Mileage, etc).

View on Github

March Madness Model

Python

Model Output:

• Constructed a Python Linear Regression Model that predicted the best metrics to predict NCAA Men's Basketball Torunament Games
• Using the NCAA App's Bracket Autopick Feature weighing the metrics the same way as the model outputed, the following bracket predicted the first 11 games correct, the first 15/16, the first 28/32 with all Sweet 16 teams alive going into the second round.
• Utilized K-Fold Cross Validation and Feature Mapping.
• Could also print direct game predictions given team statistics.

View on Github

ICU Unit Outcome Prediction System

Python (PyTorch)

• Built an end-to-end ML pipeline to predict ICU patient mortality across 12,000 patients, engineering features from 40 clinical variables using statistical aggregates (mean, median, range, std. deviation, max, and trend).
• Implemented Logistic Regression with L1/L2 regularization and RBF Kernel Ridge Regression, optimized via 5-fold stratified cross-validation with grid search over hyperparameters.
• Evaluated models across 7 performance metrics (AUROC, F1, Accuracy, Precision, Sensitivity, Specificity, Average Precision) with bootstrap confidence intervals for robust test-set estimation.
• Achieved an AUROC score above 0.85, with results visualized via confusion matrices and ROC curves.

Full-Stack Instagram Clone

Python, JavaScript(React), SQL, HTML, CSS

• Built a full-stack social media app with authentication, feed, posts, profiles, followers, explore, and comments.
• Configured 7 RESTful API endpoints and 19 dynamic routes with 5 SQL tables for data and authorization.
• Created React components for dynamic pages and static layouts, delivering fully functional features.
• Collaborated in a 3-person team using Git/GitHub, conducting code reviews and coordinating tasks.

Multithreaded Distributed MapReduce Framework

Python

• Developed a multithreaded system utilizing 4 parallel threads to coordinate distributed tasks.
• Engineered to handle 10+ concurrent worker nodes with stable throughput.
• Worked in a team of three, contributing to registration and shutdown workflows for worker management.
• Achieved sub-5-second processing times on benchmark datasets within a distributed MapReduce framework.

Search Engine With Distributed MapReduce Indexing

Python, SQL, HTML, CSS

Example Input

Example Output

• Created a search engine using segmented inverted indexing generation using a 7-Stage MapReduce pipeline.
• Implemented tf-idf text analysis with PageRank link analysis to optimize search analysis for thousands of different pages.
• Built a dynamic server-side web interface to handle user search queries and display search results.
• Collaborated in a 3 person team to properly distribute workloads.

Pipelined Processor Simulation

C

Example Input:

Example Output:

• Built a simulated pipeline processor using LC2K assembly code as input.
• Was able to have the program give all information on every cycle, listing all register and memory information.

Memory Simulation

C

Example Input:

Example Output:

• Created a program that simulates computer processor memory with given LC2K assembly code as input.
• Would note every occasion information was transferred between either the memory, cache, or processor.
• Would keep track of cache hits, cache misses, and dirty cache blocks.

SQL Clone (SillyQL)

C++

• Wrote a computer program over 1000 lines long that mimics keeping data with SQL commands.
• Included Add Table, Delete Table, Edit Table, Print Table (Both Print All and Print Where), and Join Table.
• Took Advantage of advanced C++ concepts such as utilizing both Binary Search Tree and Hash Map indexing and utilizing functors.