Building a Foundation LLM from Scratch
A blog series documenting my journey building a simple foundation language model from the ground up.
I'm Corwin Dark, a data scientist focused on applying machine learning and statistical methods to financial markets. Browse my projects below to see my work in action.
A blog series documenting my journey building a simple foundation language model from the ground up.
Using AB tests to drive greater engagement and conversion rate throughout the funnel of a game I developed on the Roblox platform, improving return by 30%.
Predicting retail trader activity through Natural Language processing across more than 700 million Reddit comments. Accomplished via PySpark, with elements of the pipeline done in both AWS and Azure ecosystems simultaneously.
Building TorchMole, a Python package for encoding atomic environments using graph neural networks.
Designing and developing a hypercasual game on the Roblox platform, focusing on engagement metrics and monetization strategies.
Can we use deep learning models to identify market conditions where particular statistical models are more effective? Improving ensemble performance through intelligent model selection.
Applying statistical forecasting methods with external datasets to predict market volatility.
Evaluating the impact of Reddit sentiment on market movements using gradient boosted trees and ensemble methods.
Building a ShinyApp web application to automate all of the rebalancing, tax-harvesting, and routine buying required to manage my ETF portfolio.
Interactive scrollytelling visualization exploring climate change impacts on Natural Bridges National Monument using D3.js.