Implemented multi-armed bandit exemplar selection (UCB, Thompson Sampling) for in-context learning with PyTorch and HuggingFace.
Achieved 85-95% recovery of baseline accuracy under non-i.i.d. conditions across diverse NLP benchmarks.
CPI: 8.2/10.0
Focus on Generative AI, Multimodal Learning, Deep Learning, and Natural Language Processing.
CPI: 7.01/10.0
Completed comprehensive AI curriculum including Machine Learning, Deep Learning, Computer Vision, NLP, and Game Theory.