Productionizing ML and the Data Science-Engineering Gap

#16 - Responsible AI and Building Trust in AI - Liu Feng-Yuan

Topic 11 of 16

Published Nov 23, 2020 · Clip 35:12–39:22 (4:10)

Feng-Yuan describes productionization as a multidisciplinary gap: data scientists lack engineering skills (containers, Kubernetes, networking) while engineers don't understand ML retraining and monitoring needs. Organizational placement compounds the problem—data scientists sit near business but must throw models over the wall to IT. He explains BasisAI's self-service approach abstracting Kubernetes while allowing framework flexibility, and insists on Git version control over notebooks for production.