Keynote Interview
summary
LinkedIn runs about 500 concurrent experiments to measure real-world impact, maintains a central data science org of ~300–350 people that embeds with product teams, and builds horizontal platforms: an experimentation platform, Pinot OLAP for low-latency analytics, anomaly detection and forecasting, a metadata/catalog system, and extensible modules like model explainability. Differential privacy is applied to ad reporting and public datasets, and adoption follows a champion adoption model with subject-matter experts and measured gains tied to experiments.
AI-written summary · about
chapters (9)
- 0:00Introduction and background
- 1:14Professional journey
- 4:40Data science at LinkedIn
- 8:03Organizational structure
- 12:20Platform and application model
- 21:02Platform extensibility
- 26:42Experimentation culture
- 35:45Differential privacy
- 45:55Leveraging data platforms
related talks
35mKeynote | Stream Processing at LinkedIn: Then and Now
34mKeynote: Building for AI at Scale: Infra, Integrity, & Innovation
16mKeynote: Marin: An Open Lab for Frontier AI
14mKeynote Analysis Day 2
17mKeynote – Aparna Ramani
31mKeynote: Become a Data Superhero How Data Can Change the World
50mKeynote Panel: AI in Finance
02h 16mKeynote Day 2 - Full