Papers.

• Show, Attend, Tell: Visual Attention for Image Captioning

  CS 4782 Deep Learning, Cornell · Spring 2025 · team of 5

  Reproduction of soft and hard visual attention for image captioning (ResNet-50 encoder, LSTM decoder), evaluated with METEOR against the original paper.

  pdfcodeproject

• Low-Level Learning

  CS 4701 AI Practicum, Cornell

  Design and implementation of C-Torch, a C++ machine learning library with numerical methods, linear algebra, distributed training and 15+ models.

  pdfcodeproject

• Voting-Based Multi-Agent Bandits

  CS 6783 Machine Learning Theory (graduate), Cornell

  Multi-armed bandit algorithms for multi-agent systems in which agents aggregate decisions through voting.

  pdf

• From Redux to Component-level State: Modernizing CMSX Architecture

  CS 4999 Independent Research, Cornell · 2024

  Migration of Cornell's course management system from Redux global state to component-level state, and its effect on developer and user experience.

  pdf