• Trained and maintained domain-specific Word2Vec models for text clustering, enabling stakeholders to proactively act on trending topics from sales-customer call transcription data (Python, SQL, TypeScript, C#)• Analyzed post-deployment performance of multiple classification models alongside business colleagues, ensuring iterative enhancements (Python, SQL)• Modernized model pipelines, boosting throughput while cutting maintenance costs (Kubeflow, Airflow, Buildpacks)• Collaborated with business experts to target high-impact problems within the sales and customer relations domains• Spearheaded machine learning projects from inception to deployment, encompassing research, data exploration, model development, and post-deployment evaluation

• Produced a survival regression model in Python to predict subscription retention across the Amazon Kids+ business, to be used as a candidate for informing multi-year financial projections. Built a testing framework to compare various modeling experiments and backtest dates, and presented the findings in technical reports to stakeholders• Deployed machine learning scientist colleague’s algorithms into reusable end-to-end systems utilizing AWS• Developed a web application prototype in Python to investigate subscription retention trends across various customer characteristics, enabling self-service analytics amongst business stakeholders• Automated the Amazon Kids+ weekly business report, reducing production time by 90% whilst enhancing iteration• Monitored and incrementally enhanced ETL pipelines critical for business reporting functions• Resolved ad hoc analyses using SQL and Python for business stakeholders, enabling data-driven decision making• Mentored non-technical business partners regarding how to use SQL to pull basic statistics and trends

• Developed a predictive model in R to forecast patient arrival counts in an emergency department as well as predict anomalous arrival surge events, empowering directors to proactively allocate resources to meet demand• Performed literature reviews and interviewed domain experts in support of model development and operationalization• Built a time series forecasting framework in R that simplifies and enables fast data exploration, model training, and model validation, thereby speeding up developer velocity for time series-specific investigations• Peer reviewed model implementation workflow proposals, regression and classification modeling code (R, Python), and machine learning infrastructure code (SQL, C#, JavaScript, Caché)• Mentored colleagues on time series analysis, anomaly detection, and predictive model development• Maintained customer relationships with analytics teams and executives as one of Epic’s data science liaisons, advocating for the customer’s desired predictive use cases and collecting feedback for ongoing investigations• Communicated analytic results to technical and non-technical stakeholders, from within Epic and from customers

Capstone project as part of the Multidisciplinary Design Program at the University of MichiganProject title: Professional Researcher Activity Analysis and Prediction• Developed a web application within a multidisciplinary team to analyze user log data of an online information service, enabling analysts to identify the cause of bottlenecks within customer workflows and act upon them

• Implemented a RESTful web service API in C# to support Apple Wallet digital cards, enabling patients to easily access their health insurance information. Integrated with the Apple Push Notification Service for real-time updates

• Developed a web application prototype in JavaScript to track the location and status of patients as well as monitor the occupancy status of rooms in a clinical setting, permitting clinicians to efficiently locate and transfer patients

• Utilized customer input to recommend products best fulfilling their needs, thus ensuring customer satisfaction and increasing sales• Resolved customer concerns with efficiency and tact, thereby increasing customer retention