Super Data Science: ML & AI Podcast with Jon Krohn 435: Scaling Up Machine Learning
Jan 14, 2021
Erica Greene, a machine learning manager at Etsy, shares her insights into scaling ML operations for their ad system. She discusses overcoming interesting failures and how her team strategically selects problems to tackle. Erica elaborates on her day-to-day tasks, the complexities of engineering at scale, and the importance of diversity in hiring. She also addresses whether data scientists really need PhDs and emphasizes the evolving skill set required in the field, focusing on practical experience over formal credentials.
AI Snips
Chapters
Books
Transcript
Episode notes
Strategic Project Selection
- Use a strategic approach when deciding on ML projects, similar to product development.
- Run a discovery process involving brainstorming, prototyping, and risk assessment.
Scaling ML Development
- Scaling ML model development involves challenges beyond infrastructure.
- Collaborative experiment tracking, shared code repositories, and canonical datasets improve efficiency.
Remote Team Management
- Erica Green prefers asynchronous communication and minimizing team meetings.
- To counter remote work isolation, pair team members on projects.
