Columbia University
Expected Dec. 2026
Operations Research Financial Engineering - GPA: 4.0
Recent Projects
June 2024 - Present
Price Prediction model utilizing LightGBM regressor based on top of orderbook feeds.
Features
- 0.52 R^2 (Strong Enough) on test dataset.
- Engineered common High Frequency Trading Labels such as Top of Orderbook Imbalance, midprice, and spread.
- Feature engineering via lag, and rolling methods.
- Generates all dataset labels in near real-time utilizing Dynamic Programming to cache frequent access results.
- Parses real world data provided by LOBSTER Data to replicate real world order intensities.
May - June 2024
Implemented the Queue-reactive model in python to simulate a real orderbook to predict probability of execution for high frequency trading (hft) algorithms.
Features
- Capable of processing 100,000 orders a second using a fast Limit Orderbook Implementation utilizing AVL Trees
- Parses real world data provided by LOBSTER Data to replicate real world order intensities.
- Vectorized all data computations to run in near-instant time utilizing Dynamic Programming to cache frequent access results
Dec. 2023 - Jan. 2024
A neural network coded in c++, utilizing no machine learning libraries.
Features
- 95% out of sample accuracy on MNIST dataset using a 3 layer neural network
- 784 input nodes, 128w hidden nodes, 10 output nodes
- Implemented RELU activation function for hidden layer, Softmax for output layer
- Cross Entropy Loss Function
- Trained on MNIST dataset