Explore industry-inspired AI projects for classroom learning, hackathons and workshops.
Ask humans to label only uncertain images and improve the model iteratively.
Adapt vocabulary exercises based on learner mistakes and spaced repetition.
Simulate changing data and retrain a model when performance drifts.
Train an agent to solve a simple gridworld and explain exploration versus exploitation.
Improve retrieval quality using user ratings, failed queries, and better chunking.
Train a small byte-pair tokenizer and visualize how text becomes tokens.
Build a small transformer encoder for sentiment or topic classification.
Train a tiny character or word-level language model for story continuation.
Fine-tune a small open model on domain FAQs and compare it with RAG.
Quantize a small model and compare speed, memory, and answer quality locally.
Train a smaller model from larger-model labels and evaluate compression trade-offs.
Generate a story outline, scene list, and image prompts for educational content.