During my high school years, I was inspired by a TED talk by Richard Turere to design a cost-effective and technically improved version of the “lion light,” a device that uses blinking lights to scare lions away from livestock, and partnered with the NGO China House for its distribution. 
It’s a game engine that crafts a navigable 2D tile-based world. The novelty lies in its use of pseudo-random seeds, ensuring each generated world is unique and tractable. The skeleton code was provided by CS61b (data structures) from University of California, Berkeley. 
It’s a Streamlit chat app allowing users to interact with PDF content, using embeddings and FAISS for efficient query-based retrieval from PDFs. 
Fine-tuning a pre-trained language model for the sentiment analysis task on the Stanford Sentiment Treebank (SST-2) dataset. The dataset is made up of sentences from movie reviews, each labeled as either ‘negative’ (0) or ‘positive’ (1). The ultimate goal was to train a model that can accurately classify a given sentence into one of these two sentiment classes.