During my summer at Hudl, I worked on the FastModel product across two migrations, one on the frontend and one on the backend.

On the frontend, I migrated over 1,300 React components from two codebases into Hudl’s primary repository. Most of the work was untangling large, monolithic Redux files and isolating legacy dependencies. To streamline it, I built a closed-loop Claude agent that used model cascading and dependency pruning to cut token use by up to 75%. Alongside the migration, I proposed and documented a new domain-partitioned Redux architecture.

On the backend, I migrated a third-party dependency that had accumulated a lot of technical debt. Working with product managers and senior engineers, I audited the system requirements and data models, which let us collapse an 18-table schema into 4 sports-agnostic tables for the initial launch. From there I set up the AWS Aurora PostgreSQL database and built the C# CRUD service layer.

Outside the migrations, I worked across the stack, improving accessibility and implementing log tracking.

During my time at FindU, I worked on the iOS app, mostly on two features: the recommendation engine and the AI advisor.

On the recommendation engine, I built a scholarship swiping interface in Swift from Figma designs and integrated Supabase event logging to track user interactions. That data, alongside 25,000+ embedded scholarships, fed a cosine similarity algorithm I wrote to recommend relevant scholarships to students in real time.

On the AI advisor, I developed a GPT-5 agent with access to 15+ functions across CRUD operations, including adding, removing, and organizing scholarships on a user’s profile. Parallel tool execution and semantic caching cut response latency by 30%.