Overview
Over Thanksgiving break of my sophomore year, I wanted to take what I had been learning in a data science course at UT and actually do something with it outside of class. I had just discovered Kaggle, was getting my first real exposure to machine learning, and movies felt like the perfect place to explore it.
So I built The Popcorn Model, a project focused on predicting box office performance using factors like budget, popularity, ratings, and runtime. No team, no deadline. Just curiosity and a long weekend.
Contribution
Machine Learning
Team
Ruthvik Jonna
Date
November 2023
Process
It started small and quickly turned into weeks of experimentation. I cleaned datasets, tested different models, and spent a lot of time trying to understand why certain approaches worked better than others.
Most of it was not glamorous. A lot of pipeline errors, a lot of retraining, a lot of improving one small decision at a time. There was no co-founder to bounce ideas off, no customer to build for. Just me, the data, and a lot of trial and error. But somewhere in that process, technical concepts that had only ever lived on slides started feeling like real tools I could actually use.
Outcome
The models were far from perfect, but that was never really the point. It was the first time I built something end to end completely on my own, and the first time I genuinely enjoyed the process of being confused and figuring it out anyway.
That mindset carried into everything that came after, Bloom included. There is just something about learning by actually building that no classroom has ever replicated for me, and I don't think that is ever going to change.