Catalog & Cocktails: The Honest, No-BS Data Podcast Shift Left Everything: How Ontology, Events, and Culture Unlocked Enterprise AI with Nachiket Mehta
15 snips
Feb 19, 2026 
Nachiket Mehta, VP of AI and Data Engineering who built ontologies and event-driven teams at Wayfair. He unpacks mapping happy vs unhappy paths, the $35M trailer story, shifting data teams left into domains, tiger teams and ops visits, and turning process maps into reusable ontologies for actionable enterprise AI.
AI Snips
Chapters
Transcript
Episode notes
Trailer Left In Yard Became $35M Problem
- Nachiket describes Wayfair's discovery that KPI mismatches came from ambiguous event and identity definitions, not just metric definitions.
- He recounts hiring an ontologist and mapping supply-chain events to reveal real-world mismatches that broke forecasts.
Culture Beats Pure Tech Investment
- The technical foundation is only ~20% of the solution; change management and culture are ~80%.
- Ontology work pays off only when paired with organization change and executive sponsorship.
Map Happy And Unhappy Paths First
- Map happy-path and unhappy-path events end-to-end by visiting operations and building process maps with operators.
- Use those maps to connect events to systems and detect anomalies that lead to business cost.