- Collaborated with a team of four students and a Google mentor on building a health application that takes live camera feeds of food to be tracked via its nutritional information - Took advantage of MobileNets and computer vision to create performant, yet accurate models that can predict even branded foods and get the right nutritional information - Took up the role of project manager and presented the final application in a live demo to other mentees and Google engineers at the end of the program.

- Developed a three-month-long curriculum to onboard new members on Java programming, control theory, and good software engineering practices, used and cited by hundreds of people since.- Managed our team of 40+ software engineers, allowing for streamlined development and testing of robots following the AGILE methodology to ensure desired features were implemented- Headed software-mentoring efforts for the team, hosting hybrid education sessions and in-competition aid to other FIRST teams, even internationally to countries like Turkey, Taiwan, etc.

- Led subteams in programming a swerve drivetrain, double-jointed arm, and other subsystems for three world-championship-qualifying robots; ranked 24th, 26th, and 86th in the world during the '22, '23, and '24 seasons, respectively.- Designed and single-handedly rewrote the overhaul of the team's LED subsystem, allowing for clear robot-driver communication. - Researched and worked on advanced robotics projects such as robot vision detection with convolutional neural networks, ROS integration, and a turret subsystem.

- Underwent onboarding training on web scrapping and data engineering, practicing by extracting and analyzing commercials trends, news media, and job data from websites using Selenium, urllib, and bs4 to be used for real-world situations and hiring solutions used by the company.- Built a job aggregation site with over 60k jobs from job sites (Indeed, ZipRecruiter, etc) using full stack technologies

- Collaborated with a team of four students and a Google mentor on building a supervised multi-modal machine-learning algorithm to categorize and generate science word problems as a study tool for students - Used techniques such as vectorization, multinomial naive-bayes, and gridsearchcv from the sci-kit-learn library to create and optimize our model, achieving an 83% precision weighted average.- Responsible for building the final web app; presented in a live demo to other mentees and Google engineers at the end of the program.