Scott Davis Email & Phone Number

Carrboro, Nc, Us

- Integrated AWS and Databricks instances to allow for Spark processing of billions of rows of TV viewership data housed in parquet files, greatly reducing data retrieval time.- Led a data science team of five members in setting priorities and roadmap, mentored other data science members, and directly managed MVP intern program.- Identified issues with extrapolation methodology and created an automated QA process based on computer vision outputs to improve accuracy five-fold.- Defined production machine learning model drift best practices and API endpoints for accessing models, including a standardized data visualization dashboard in Azure Databricks.- Created a natural language processing machine learning model using social media data from teams and athletes to determine brand sponsors with 80% accuracy.- Disaggregated viewership data to trace the habits of households that watched Formula1 and NASCAR races over the course of a week in order to create viewer personas to serve as the basis for a marketing piece.

- Created and maintained multiple machine learning algorithms on AWS EC2 to solve self-identified business problems, including a regression model that reduced data delivery time by 50%.- Cleaned, structured, and analyzed large sets of social media and broadcast viewership data in Spark totaling more than 100 million rows to enable cross-functional productivity. - Managed migration of data science development processes and data warehouses to Azure Databricks.- Extracted data from 4 disparate sources for custom client applications, enabling sales to sign contracts worth $1.8M annually.- Used NLP and cosine differences to determine similar social media posts across platforms, saving 60% QA time on average.- Designed multithreaded production code to speed up creation of video thumbnails from days to minutes.-Idealized data-based marketing initiatives to expand company capabilities beyond standard offerings allowing for upsell opportunities to customers.

- Applied machine learning techniques to improve previous models for object detection. Real objects are now detected with 35% greater accuracy, with minimal gain in false positives.- Designed and wrote successful grants for highly competitive international telescope time (James Webb Telescope), including funding support of over $1M.- Advised graduate student projects, led group meetings, and assigned research topics based on student availability and experience.

Tallahassee, Florida, United States

- Performed data reduction and analysis on a large sample of supernovae, including principal component analysis, leading to the creation of standardized tools for astrophysical analysis.- Developed SQL backend and JavaScript/PHP frontend for collaboration website.- Created criteria for further classification of objects post-machine learning for specific use cases.

Clemson, South Carolina, United States

- Performed data reduction for research done on planetary formation. - Modeled astrophysics data using the IDL programming language. - Successfully converted IDL programs to Python.

Doctor Of Philosophy - Phd, Physics

Master Of Science - Ms, Physics

Bachelor Of Science - Bs, Physics