Projects

Supply-Chain Risk & Financial Markets

Independent Research | 2026 - Present

• Built a Python pipeline to collect and analyze SEC 10-Ks, XBRL, and equity-market data for 450 public companies in nine sectors.
• Used Word2Vec to create supply-chain risk and resolution dictionaries from 20 years of SEC filings and earnings calls.
• Applied FinBERT to quantify information in corporate disclosures and investigate relationships with market returns.
• Built event-study and regression models comparing cumulative abnormal returns (CAR) with expected returns using a four-factor benchmark and firm-specific market-response patterns.

Machine Learning Reproduction & Experimental Projects

Independent Projects | 2026 - Present

• Reproduced optimization experiments from the Adam paper (Kingma & Ba, 2015), implementing Adam, AdaGrad, RMSProp, and SGD/Nesterov and comparing convergence behavior on neural-network training loss.
• Implemented and evaluated neural-network architectures using NumPy and PyTorch, including MLP and convolutional models on MNIST/CIFAR-10.
• Designed controlled experiments to investigate optimizer hyperparameters, convergence behavior, dropout, and model architecture.

Technical skills

Programming: Python (PyTorch, NumPy, Pandas, Scikit), Matlab
Machine Learning: NLP, transformers, neural networks, reinforcement learning, optimizer tuning, supervised learning, PPO/GRPO