[2022] Wordle Solver: Tool that solves Wordle in 3.63 guesses on average. Calculate the distribution of letters per-position in the corpus. Choose the candidate that maximizes the joint probability. [Colab]
[2021] Fantasy Basketball Optimization: Tool that formulates Fantasy Basketball 9CAT lineup selection as an Integer Programming Problem. Allows users to specify a list of players and can then solve the corresponding optimization problem. [Colab]
[2019] Approximate Near Neighbor Seach under l-infinity: Covers Indyk's data structure for l-infinity approximate near neighbor search (c-ANN). Proposes two alternative data structures for c-ANN under l-infinity for low-dimensional spaces. Also covers hardness of c-ANN under l-infinity for c < 3. [PDF]
[2019] Identity Testing: Survey on sublinear algorithms for both identity and uniformity testing. Includes upper and lower bounds for collision-based testers, coincidence-based testers, and instance-optimal testers. [PDF]
[2018] Deep Learning for Network Traffic Classification: Covers predicting Server Name Identification (SNI) using HTTPS features (packet sizes, payload sizes, inter-arrival times, directionality). Compares machine learning and deep learning classifiers such as Random Forest, CNN, RNN, CNN-RNN, and Ensemble. [PDF] [Poster] [Repo]
[2017] Microbiome Clustering: Uses unsupervised machine learning techniques to better understand and describe microbiome samples. Advised by Professor Itsik Pe'er at Columbia University. Published in Journal of Healthcare Engineering, vol. 2017. [PDF] [Repo] [Video]
[2017] First-Order Optimization Methods: Survey on properties of a variety of first-order methods, including naive and accelerated gradient descent, mirror descent, and stochastic gradient descent algorithms. Includes descriptions of the classical algorithms as well as breakthroughs (Katyusha, variance reduction, linear coupling, momentum). [PDF]
[2017] Faster Linear Programming: Survey on methods for fast linear programming, focusing predominantly on the evolution of interior point methods in the last thirty years. [PDF]
[2017] Alarm Fatigue: Explores the issue of alarm fatigue in physiological monitoring systems in hospitals, and to approach the problem of alarm fatigue from the perspective of machine learning. [PDF]
[2016] TeapotCollector5000: Final project for Computer Graphics, made with OpenGL and GLUT frameworks. [Repo]
[2015] Linkedin Scraper: Uses Google Search and Selenium Webdriver to find LinkedIn profiles matching desired key words. [Repo]