How Mars' 12-person data science team powers AI across a $50B org
rschmelzer · x · 2026-09-16
On Small Team Big Scale podcast (EP045), Shashank Kadetotad, Global Sr. Director of Enterprise Data Science and AI at Mars, explains how a 12-person team serves 100,000+ employees and hundreds of brands across a $50B conglomerate.
Key takeaways:
- Composable "accelerator" modules: 15 reusable Lego-block-style modules (forecasting, optimization, etc.) cut deployment from weeks to days
- Governance as an enabler, not a bottleneck: standardized governance actually speeds replication across business units
- AI Lab with upfront failure criteria: a safe experimentation space where pilots define what failure looks like before starting, so weak projects get killed early
- Amazon-style pizza-team principles and vendor-agnostic stance amid a shifting AI landscape
Covers supply, demand, finance and HR use cases — practical lessons for lean enterprise AI teams.
More from Companies & People
- Reporter probes whether AI school Alpha School is a cult, publishes insider email — benjaminjriley · 2026-09-16
- Ethan Mollick: AI is blurring coding, design and PM roles inside companies — emollick · 2026-09-16
- Deep profile of AI detector Pangram: how it was built, whether it works, why it matters — TuhinChakr · 2026-09-16
- AI semiconductor endgame: non-zero inference costs keep the buildout running longer than the internet's — BenBajarin · 2026-09-16
- How tweeting through SB 1047 debates took an EleutherAI researcher to a job at METR — CFGeek · 2026-09-16
- OpenAI launches Codex Ambassador Program for power users — paw_lean · 2026-09-16