I'm a Machine Learning Engineer based in Los Angeles, CA, with a unique background spanning a decade in power systems engineering. My professional journey began with building and maintaining the critical network simulation models that ensure grid stability for major utilities.
This deep experience with complex, real-world systems inspired me to pursue a Master's in Computer Science from Georgia Tech with a specialization in Machine Learning. I am passionate about applying my advanced ML skills to solve high-stakes problems in the tech and finance sectors, particularly in areas like energy trading, predictive modeling, and building intelligent automation pipelines.
- 🔭 I’m currently focused on completing my graduate studies and building a portfolio of ML projects.
- 📚 I'm deepening my expertise in predictive modeling, reinforcement learning, and computer vision.
- ⚡ In my free time, I enjoy exploring workflow automation tools and tinkering with game development in Unity.
My primary skillset is centered around Python for machine learning and data science. I am experienced in the end-to-end development lifecycle, from data analysis and model training to deployment. My engineering background provides me with a strong systems-level approach to problem-solving.
While much of my recent academic and professional work resides in private repositories, below is a summary of key projects and experience.
- Built a data processing pipeline to take 15 years of raw OHLCV data and reconstructed features out of it to predict price breakouts of 8% or more
- Used LightGBM as the backend, created a genetic algorithm to use for feature selection after generating over 100 different technical indicators to determine which feature combinations have highest information gain and feature importance
- Built a market simulator to handle futures price expiration and hand off between months to train effectively on back-adjusted futures data
- Edge over market is over 15% greater than randomized guessing with a heavily class skewed model
- Used data downsampling to deal with overwhelming class majority of the dataset being 'hold' or 'nothing' because of the nature of searching for black swan events.
- Applied randomized optimization algorithms like Simulated Annealing and Genetic Algorithms to find high-quality solutions for computationally-intensive problems.
- The process involved implementing the algorithms, collecting performance metrics on solution quality and convergence speed, and creating figures to compare their effectiveness.
- This work was summarized in a comprehensive report, demonstrating the application of different optimization heuristics for solving complex problems.
- Designed, built, and maintained large-scale power flow simulation models for a major utility's Energy Management System (EMS).
- These models were critical for running a state estimator used by system operators for real-time grid analysis, contingency planning, and ensuring operational stability.
- Developed a deep expertise in analyzing complex systems and managing mission-critical data pipelines.
- Developed and backtested several trading strategies based on machine learning principles.
- Implemented algorithms to analyze historical market data, identify predictive signals, and generate trading orders.
- Gained experience with financial data analysis libraries in Python and frameworks for evaluating model performance in a trading context.
- Built and trained a Convolutional Neural Network (CNN) from scratch to classify handwritten digits from the MNIST dataset.
- Engineered a model that successfully achieved 97% accuracy on the test set.
- This project involved data preprocessing, model architecture design, hyperparameter tuning, and performance evaluation.