Cuffed
Building a dating algorithm. Work done during Summer 2024 machine learning engineering internship with Cuffed Dating
Project overview:
- Trained and evaluated several multimodal deep learning models to predict if a user is likely to swipe left/right on another user for improved deck creation using Python, NumPy, pandas, and PyTorch.
- Increased prediction accuracy by 64% and precision by 167%, when compared to naive approaches.
- Built off of Gale-Shapley algorithm for optimal matching and deployed the model for deck generation using Python, TensorFlow, and AWS SageMaker.