• Implemented hardware for full body motion capture and software for data processing.• Developed Python scripts to optimize processing load and ensure smooth app performance.• Designed code for live audience text message processing during a tech panel at New York Tech Week.

Designed and implemented an advanced model to optimize the matching of requests with available packages, enhancing operational efficiencyUtilized K-means clustering to categorize and analyze existing data sets, enabling more accurate data-driven decision-makingStandardized the processing of incoming request data, ensuring consistent and efficient alignment with identified data clustersEmployed grid search methodology to fine-tune hyperparameters, significantly improving model performance and accuracyIdentified and removed variables with high multicollinearity, enhancing model accuracy and preventing overfittingDeveloped an advanced predictive model to forecast Set Top Box reboots, significantly enhancing system reliability and user experienceImplemented a sophisticated Convolutional Neural Network (CNN) featuring two hidden layers, demonstrating expertise in deep learning techniquesUtilized the Adam optimization algorithm for stochastic optimization, achieving an exceptional model accuracy of 96%.

Worked on a small team as the only data scientist. Created and maintained the project database. Initial database contained many categorical features where the significance of each value was unknown. Through a combination of identifying the industry standards and reading product operation manuals these values were found and assessed for their usefulness to the project. Unused features were removed and the remaining column values standardized. Used Python to geocode entire database and added lat/long coordinates. Also created weekly rolling averages for multiple variables using hourly time series data. Created sanitized subsets of that database that could be shared with other businesses.

Developed and presented the team's data visualization. Created time series graphs for thousands of sensors over a time span of several years, as well as visualizations of all 25 million observations. These graphs were used to identify outliers and investigate trends. Enhanced the efficiency of the code by making adjustments to reduce run time and to prevent low density graphs. Most data visualization was done using R. Created PDF and slide decks using R's knit function in order to present and share findings with the team.

Worked as a team member to set-up and execute firework displays.Coordinated with other Tri-State teams to deliver consecutive displays.Trained new employees in execution and safety standards.