Data Scientist
Current• Led development of a significant model update in PyTorch converting a binary model with ambiguous classes to a multinomial model that better fits the current and future use case while increasing the ability to test the model and report KPIs.• Collaborated with team members and product team to develop a method for ordering the output of an anomaly detection algorithm based on uncertainty for a better customer experience.• Developed KPI’s for a key model that reflect downstream impact… Show more • Led development of a significant model update in PyTorch converting a binary model with ambiguous classes to a multinomial model that better fits the current and future use case while increasing the ability to test the model and report KPIs.• Collaborated with team members and product team to develop a method for ordering the output of an anomaly detection algorithm based on uncertainty for a better customer experience.• Developed KPI’s for a key model that reflect downstream impact which included developing a model to reduce the amount of data to be labeled and coordinating with a third party company for labeling.• Audited internal training data to understand the scale and impact of poor training samples on our model, which led to an effort to monitor new and existing training data.• Implemented content similarity method using Word2Vec embeddings in SQL to improve robustness of internal topic similarity process and reduce onboarding time for new employees.• Coached new and existing employees on SQL and Google BigQuery, the structure of our databases, and how to find necessary data. Show less
