Henry Corkran

About

  • Hello! I am a software engineer working on an MS at UPenn. I was a trader and derivatives analyst on Wall Street for several years before moving to tech after becoming enamored with the ongoing revolution in deep learning and the monumental systems powering it.
  • I have experience building a diverse set of software, including real-time financial analysis systems, databases, web crawlers, and fullstack apps.
  • ML is now my primary focus, and I've designed and implemented deep models for fields ranging from biology to board games.

Professional Experience

    SMBC Capital Markets - Trading Analyst
  • Began as a junior trader at one of the world's largest banks and found a passion for software development and data science.
  • Built dozens of tools and automations with Python, C++ and SQL to integrate the firm's databases and valuation systems and speed up the pricing, execution, and booking of trades.
    Rosenblatt Securities - Equity Research Intern
  • Conducted research for a combined universe of 40+ stocks in semiconductor, fintech, and digital asset sectors.
    Group One Trading - Options Trading Intern
  • Trained in trading and options theory at a leading market-makers in exchange-listed derivatives.

Projects

    DeepOthello - AlphaGo for Othello
  • AI board game opponent powered by neural net architecture inspired by DeepMind's AlphaGo and AlphaZero.
  • Trained in PyTorch, game logic and interface written in Java.
    Protein Forge
  • A novel softmax-based pseudo-alignment model for predicting how multiple 3D protein structures chain
  • Incorporates feature engineering into existing protein foundation models to reduce time on MSTA search tasks by several orders of magnitude versus classical techinques
  • Interpolated functional JAX training scripts to more performant PyTorch variants
    European Defense NLP Research Database
  • Web crawling pipeline built for a Penn professor's research to extract and organize reports on the Russo-Ukrainian war from top European defense think tanks. Utilized MongoDB, python, and LLM tooling in the pipeline.