Hello, I'm
Patrick Melan
Programmer.
Software engineering student at Drexel, building full-stack web apps and machine learning tools with TypeScript, Python, and AWS.
2025 Graduation Speech | Click here to watch
About Me
Hello! I'm Patrick Melan, a sophomore at Drexel University studying software engineering. I enjoy building full-stack web apps and useful data tools & dashboards, mostly in TypeScript and Python. My stack usually involves Next.js, Python, and some combination of AWS services, but I'm always trying to learn new tech.
Outside of code, basketball takes up a lot of my time. I am a PIAA basketball official, officating varsity games throughout the Lehigh Valley and Philadelphia, and also a student manager for the Drexel Women's Basketball team, doing things from traveling with the team to away games to helping record statistics at practice.
Recently, I have been spending most of my coding/research time on machine learning, learning the funamentals of building and training models using different statistical models such as linear regression, decision trees, and neural networks. While I would consider myself at a basic-intermediate proficiency, I plan on continuing my learning and understanding throughout this summer.
Relevant Experience
Freelance Software Engineer
FreelancePolity Inc.
May 2025 - Present
Built a multi-client government data review platform in React + Tailwind CSS, serving 9 municipalities through a Clerk-authenticated multi-step wizard with two core flows: field mapping and record validation.
Key Achievements
- •Designed and deployed an Express REST API on AWS EC2 integrating S3 file uploads, DynamoDB batch writes, and Athena SQL queries to store and surface permit, GIS parcel, and code violation data at scale
- •Implemented server-side fuzzy matching with Fuse.js over paginated Athena results to detect MAT vs. GIS parcel discrepancies (addresses, owner names) across thousands of government records
- •Bundled the React app with custom Webpack config and deployed to S3 for embedding in Polity's Webflow site
Technologies Used
Projects
NCAA ML Bracket Predictor
Data ScienceXGBoost model trained on Kaggle NCAA tournament data augmented with Torvik adjusted efficiency ratings to predict game outcomes. Runs 10,000 Monte Carlo simulations of the 68-team bracket to generate per-team round-by-round advancement probabilities, then surfaces results in a Next.js frontend with live betting line comparison via the Odds API.
Easton Flipped
Client WorkFull-stack real estate platform for a property renovation business, built with Next.js 15 and PostgreSQL via Drizzle ORM. JWT authentication is enforced at the Edge via Next.js middleware, protecting an admin dashboard for managing property listings with S3 image uploads and four distinct contact form flows (sell/buy/general/quick) with reCAPTCHA v3. Deployed as a multi-container Docker Compose stack with Caddy (auto HTTPS), self-hosted Umami analytics, and DNS via AWS Route 53.
Polity Data Request Form
Client WorkFull-stack multi-municipality data review platform built for Polity Inc., a government data consulting company. Clients authenticate via Clerk and walk through guided workflows to either map and request specific data fields or validate existing government records (permits, code violations, GIS parcels) before a data migration. An Express API on EC2 backs AWS Athena SQL queries over S3-stored data and fuzzy-matches MAT vs. GIS parcel records with Fuse.js, surfacing discrepancies in a multi-step React wizard.