Lean AI Massive Scale: How a 12 Person Team Powers a $50 Billion Dollar Company

Supported by qBotica (https://qbotica.com) A 12-person data science team driving AI transformation across a $50B+ global conglomerate? That's exactly what Shashank Kadetotad is doing at Mars. As Global Sr. Director of Enterprise Data Science and AI, Shashank leads a focused team that builds composable "accelerator" modules — reusable AI building blocks that have compressed deployment timelines from weeks to days. He shares how the team defines failure criteria before starting any pilot, why governance is their biggest enabler of scale, and how the AI Lab they built provides a safe space for experimentation with hard boundaries. Whether you're running a startup or an intrapreneur inside a massive enterprise, the lessons from this conversation are universal: systemize everything, stay vendor-agnostic, and don't fall in love with your product. 🔑 Key Takeaways: • 12 people serving 100,000+ across supply, demand, finance, and HR • 15 reusable "accelerator" modules that cut deployment from weeks to days • Governance as your biggest scaling lever, not a bottleneck • The AI Lab model: safe space + mandatory failure criteria • Pizza-team principle: no more than 3-5 people per meeting • Stay vendor-agnostic to maintain strategic flexibility • Define criteria for failure, not just success • Systemize everything — if you can't systemize it, question whether it's worth doing ⏱️ Timestamps: 00:00 — Opening 01:44 — Introduction and background 04:54 — The 12-person team driving enterprise AI at Mars 07:38 — Not being the bottleneck: harnessing the organization 09:24 — Accelerators: composable AI Lego blocks 12:40 — From weeks to days: deployment compression 15:17 — Governance as an enabler of scale 18:01 — Playbooks, shadowing, and production-ready AI 22:07 — From POC to production: the real effort 26:20 — Tool sprawl and the build-vs-buy decision 31:22 — Adaptable systems and tying everything to value 35:47 — Building the team: ownership and the AI Lab 38:41 — The AI Lab: experimentation with failure criteria 42:11 — Rapid Fire: one tool, one process, one tip 49:14 — Systemize everything 50:05 — Closing and subscribe 🔗 Links: • Mars: https://www.mars.com • Shashank Kadetotad on LinkedIn:   / shashankkadetotad   • Small Team Big Scale Podcast: https://smallteambigscale.com/podcast