Data Scientist
Current- Upgraded support ticket classification software to utilize Google’s PALM2 llm via Vertex AI. ~2,000 Salesforce support tickets per week are analyzed for sentiment, key issues, and resolutions via custom python code. Resulting data arE stored in GBQ and made accessible for teams on superset and sheets.- Helped research and prototype modern customer chatbot solutions from google, chatgpt, and open source models- Maintained monthly automated reporting of customer support slides to resellers/resellers- Assisted with feature development in company-wide customer churn prediction model using GBQ and python. Predictions from this model are used by teams to triage potential churn before it happens. - Maintain airflow DAG for customer churn prediction and reporting. 10,000 customers have their churn risk and top 3 reasons predicted each month using a scikit learn random forest model.