Abstract: With only 13% of data science projects ever making it into production, transforming computer vision projects from costly experiments to drivers of real business value is a key challenge organizations face today. In order to accelerate the delivery of vision AI, innovation leaders must learn to expand the perimeter of AI development by empowering subject matter experts and those closest to the problem to directly contribute alongside data science teams to the success of AI projects. This talk reflects on CrowdAI’s experience working with leading government and commercial partners and how we believe the modern enterprise can rapidly scale computer vision models that are sensitive to granular, pixel-level quality, safety, and operational issues.
Bio: Marcella Valentine is a founding technical program manager at CrowdAI. She serves as a thought leader on building the right data foundation for scalable machine learning projects, from the United States Government to the Fortune 500, turning customer ideas and needs into actionable machine learning project plans. Her background is in large scale geospatial projects and has previously worked in groundwater flow research and contamination cleanup. She holds a Bachelors of Science in Geophysics and Geographic Information Systems from the University of California, Davis.