A full stack two-sided marketplace built from scratch around big data, web scraping, ecommerce and modern UX.

In the summer of 2022 I was contacted by a recruiter to take a project from a rudimentary proof of concept to a fully functioning MVP. The client was a startup trying to capitalize on the new opportunities NCAA athletes have to get paid and establish brand partnerships.

The goal was to build something that combines the two-sided marketplace features of an app like Cameo, Turo or Upwork and the discovery tools that rely on big-data to find trending, hot or niche matches for a business to an athlete.

Businesses, agents, or athletes can come to the app, sign up and track metrics coalated to a bunch of individual scores generated by a slurry of different factors. All of the various score elements go through an algorithm to then produce a complete summary score. As the application grows the hope is to continually include more factors to create the underlying scores.

The team came to me as their founding software developer had to step down to pursue other opportunities and wanted me to take what was then several microservices and help combine them into a fully function ecosystem, all while developing the product at the same time.

Backend / Data Development

I decided to start from scratch, taking the data models assembled in a Django application as a base. The first task was to web-scrape every college athlete that existed and gather any relevant metadata pertaining to them. All of this logic was then put into lambdas to continually get new players as they came into the NCAA. We also had to develop pipelines that continually track the performance, engagement and sentiment of any relevant athlete.

Tools Used:

• Hasura
• Postgres Database
• GraphQL
• Google cloud functions
• Python/Flask for data scraping

eCommerce Development

After the initial workflow for tracking and rating athletes, we needed two create the two-sided marketplace. This required a dashboard for both athletes and companies to create and track potential partnerships.

I started by working alongside the leadership team to generate a variety of user stories and epochs corresponding various commerce workflows. We then set up an inventory of products and variants that could potentially be purchased. I structured this data after my experience using the shopify API.

The price of each "product" is dynamic meaning that it corresponds to an algorithm based on the data of each player. Each item needed to be calculated from this formula and then cached so that it could be accessed quickly.

Tools Used:

• Redis
• Postgres
• Google Sheets API
• Figma / Material UI