Computational Fluid Dynamics (CFD) R&D

● Created a code generator using libclang to generate C++ code for classeswith annotations in the Snapchat Lens Engine. This follows a similardesign paradigm to Unreal Engine’s C++ blueprint annotations.● Made the Snapchat Physics Engine more cross-platform deterministic bychanging the pseudorandom number generation algorithm● Helped migrate the Snapchat Lens Engine to use Metal for iOS● Created an OpenGL graphics pipeline cache to improve shader loading time● Operational C++ experience through on call

Developed an NLP model to analyze customer reimbursement issues.

Developed TSPH (Trick Smoothed-Particle Hydrodynamics), a fluid simulation engine within the NASA open source Trick simulation framework, which allows for data recording, the ability to change simulation parameters during runtime, and the ability to control the time evolution of the simulation. -->Used NVIDIA CUDA GPU hardware acceleration to improve performance of simulation-->Used Marching Cubes algorithm to render a closed form mesh of the fluid rather than discrete particles-->Researched various communication schemes between the simulation backend and the OpenGL graphics frontend including TCP, UDP, and shared memory-->Developed an OpenGL graphics frontend for the simulation-->Potential applications in the NASA Artemis Program for simulating fluid in tank of rocket, air interactions with spacecraft, and splashdowns of spacecraft in ocean

Developed an Anomaly Detection System for Reimbursement Policies in Python (PySpark) and Typescript (AWS CDK)-->Expected to save on the order of $10 million per year by alerting on anomalous policy behavior-->Created graphs, anomaly detection models, and dashboards of critical metrics in AWS CloudWatch based on reimbursement decision logs-->Setup alarms to activate if certain metrics were outside the predicted range of the anomaly detection model-->Created peer-reviewed High-Level Design, Low-Level Design, and Documentation

Developed a test data lookup tool in C# and Microsoft SQL Server-->Able to easily find how many 5G radios were falsely reported as defects. -->Able to look at a complete failure history for a specific serial number that was falsely reported. -->Tool cut the process down from 120 minutes to 5 minutes.Developed an Excel to XML conversion tool in C#-->Able to convert test engineering documentation into test automation project files.-->Eliminated risk of data conflicts between two files.-->Documentation became an accurate representation of what is being used in production.Developed an XML to Excel conversion tool in C#-->Able to convert test project files into a presentable format for the R&D team. -->Tool cut the process down from 90 minutes to 30 seconds.-->Eliminated risk of data conflicts between two files.Developed a Python script to populate Excel macro used for statistical process control calculations -->Reformatted test data into a usable format for statistical process control calculations.

Developed a Java program that parsed company data and produced a data visualization and a CSV report from that dataData visualization showed specific client’s uptime duration and data usage through bar graphsProgram processed large datasets (tens of thousands of logged entries) efficiently

Developed a Java program that read barcode scans from a scanner and updated company’s inventory database Company was able to effectively manage inventory after program completionCollaborated with coworkers in order to determine the company’s needs for the programCreated detailed documentation on how to use the program after project completionPresented a demo of the program in front of several coworkers and managers