Discrete Structures (CMSC250)• Led weekly lab sessions and recitations for 50+ students, covering topics such as propositional logic,induction, contradiction proofs, probability, and counting techniques.• Graded exams, provided feedback, and tutored 100+ students to improve academic performance by 10%.

• Engineered deep neural networks and partial differential equations to model COVID-19 spread across the world using 10000+ instances of official CDC and WHO health data to create an official UMD dashboard.• Yielded 90% accuracy on historical testing data with current algorithms.• Constructed an online dashboard that includes predictions for future malaria outbreaks using Pandas, Flask, TensorFlow, Selenium, AWS, Firebase, and ArcGIS, available to 10000+ global users.

• Architected various deep learning algorithms, recurrent neural networks, and statistical models to generate sector weightings on an S&P500 portfolio, yielding high accuracy on historical backtesting data.• Utilized QuantConnect, TensorFlow, Pytorch, Pandas, NumPy, and Matplotlib to analyze data and develop statistical prediction models.• Constructing large language models to perform sentiment analysis on online news articles scraped using Selenium and Twitter and Reddit posts scraped using the respective APIs with the goal of predicting future stock movements.

• Developed 8+ Deep Neural Networks and Convolutional Neural Networks to identify explosives, chemical warfare agents and drugs of abuse by diffuse infrared reflectance spectra.• Proposed novel signal accentuation process that improved the ability of machine learning algorithms to classify data involving discrete or continuous signals by at least 10%, cited by 10+ other researchers.• Leveraged Pandas, TensorFlow, and various other Python libraries to visualize and analyze InfraredBackscatter Imaging Spectroscopy (IBIS) hypercube data for pixel-by-pixel spatial and spectralclassification of chemicals within a hypercube image, increasing accuracy to 99%.