Abstract: The gap between applying data science and studying data science is huge. Data drift, cold starts, sudden scaling, and competing priorities are just a few challenges that face data scientists on the job but are rarely encountered while learning.
In this lightning overview, we'll discuss the most impactful changes you can make to your data science practice. Topics include running a model in shadow mode, data versioning, estimating costs, and communicating impact to a non-technical audience.
After this talk you will learn how to
- Assess data quality for use case
- Build better training data sets
- More accurately estimate live performance
- Translate model performance to business performance
- Get a promotion
Bio: Kerstin is CEO and Co-founder of SuperUse, a collaboration platform. She has led data science initiatives at startups across industries, from healthcare to CPG. She takes pride in mentoring fantastic data scientists and nurturing talent. A builder at heart, she regularly pushes code, trains models, and uncovers insights. She has Masters degrees in Mathematical Computer Science and Mathematical Statistics. She is expecting her PhD from Cornell in early 2023. She spends her free time going on long hikes with her two small dogs through the big mountains outside Seattle.