Planned, designed, and built an AI inference pipeline for ingesting raw images, performing AI inference, and returning the produced data to Postgres RDS database. Built with scalability and automation in mind, leveraging containerization, AWS Lambda, EC2, and S3. Responsible for training and maintaining all AI models used to produce client-facing data. Primary focus on computer vision classification and object detection. Leveraging state-of-the-art open-source architectures such as… Show more Planned, designed, and built an AI inference pipeline for ingesting raw images, performing AI inference, and returning the produced data to Postgres RDS database. Built with scalability and automation in mind, leveraging containerization, AWS Lambda, EC2, and S3. Responsible for training and maintaining all AI models used to produce client-facing data. Primary focus on computer vision classification and object detection. Leveraging state-of-the-art open-source architectures such as YOLOv8 and SAM, as well as developing proprietary architectures with Pytorch. Planned for, developed, and trained all production AI models; deployed to both cloud and edge for inference.Responsible for gathering, cleaning, and maintaining raw data to be used in AI model training. Manages relationship with team of labelers. Custom-built a synthetic data generation pipeline by leveraging generative AI. Increased size of in-house datasets 12x over in less than 1 week at minimal cost; resulted in immediate 31% increase in F1-score of the relevant model. Trained numerous classification models to perform at near-human accuracy (99%+) in rapidly changing production environments without consistent backgrounds or attributes. Show less

Led detailed analytic project to provide greater insight into organization's marketing funnel. Utilized XGBoost, survival analysis (log-rank test, Kaplan-Meier Estimator), and panel data approaches to predict the likelihood of a deal closing won. Subsequently developed Tableau dashboard to dynamically track funnel at a given point in time, make sales and marketing data transparent at varying levels of granularity, and refine quarterly forecast projections. Presented findings to COO, CMO, and… Show more Led detailed analytic project to provide greater insight into organization's marketing funnel. Utilized XGBoost, survival analysis (log-rank test, Kaplan-Meier Estimator), and panel data approaches to predict the likelihood of a deal closing won. Subsequently developed Tableau dashboard to dynamically track funnel at a given point in time, make sales and marketing data transparent at varying levels of granularity, and refine quarterly forecast projections. Presented findings to COO, CMO, and other executive leadership, and further discussion yielded marketing and sales operational changes prompted by this analysis. Show less

Analyzed organization's support tickets using NLP approaches (Python's scikit-learn and spaCy) to identify recurring themes and opportunities for client self-service. Presented results to department leadership; directly led to the creation of dozens of client self-help articles, reducing number of tickets-per-client by 8% and allowing Support team to focus attention on higher priority tickets. Developed Tableau dashboard for Support team leadership to track team KPI's and individual performance… Show more Analyzed organization's support tickets using NLP approaches (Python's scikit-learn and spaCy) to identify recurring themes and opportunities for client self-service. Presented results to department leadership; directly led to the creation of dozens of client self-help articles, reducing number of tickets-per-client by 8% and allowing Support team to focus attention on higher priority tickets. Developed Tableau dashboard for Support team leadership to track team KPI's and individual performance on a dynamic basis. Dashboard serves as executive leadership/C-team's primary window into team's success. Codified, cleaned, and optimized routinely-run Redshift SQL queries as views and materialized views for efficiency and consistency across Tableau dashboards. Show less

Utilized Python to interact with third-party APIs for data collection; created interactive executable for team members to painlessly automate routine data reporting. Created Excel dashboard and presented to upper-level management/C-team. Used Python's Pandas and nltk libraries, among others, to dissect trends in quarterly client satisfaction surveys and present recurring pain points. Identified need for additional analytics reporting tools: led effort to install Google Analytics and FullStory… Show more Utilized Python to interact with third-party APIs for data collection; created interactive executable for team members to painlessly automate routine data reporting. Created Excel dashboard and presented to upper-level management/C-team. Used Python's Pandas and nltk libraries, among others, to dissect trends in quarterly client satisfaction surveys and present recurring pain points. Identified need for additional analytics reporting tools: led effort to install Google Analytics and FullStory on new client site. Show less

Virtual project-based data analyst experience responsible for deriving meaning from 20,000+ spreadsheet transactions and delivering meaningful insights to medium-sized corporation. Transformed data using Excel's PivotTables/ VBA and visualized trends with Tableau software. Communicated observations through email and concise PowerPoint presentations. Note: Not an official KPMG employee/ intern- participant through InsideSherpa