Building a retrieval-augmented chatbot for small-business information and exploring where RAG helps—and where it does not.

Fine-tuning a small language model on Snoop Dogg and Shakespeare to explore data preparation, tokenization, and causal language modeling.

Fine-tuning a pretrained Vision Transformer on a 6,000-image subset of Fashion-MNIST, with a custom PyTorch training loop and a macro F1 score of 0.9318.