From POC to Production: How Mondelēz Builds Trustworthy AI скачать в хорошем качестве

From POC to Production: How Mondelēz Builds Trustworthy AI

In this episode, host John Russo speaks with Nitin Gupta, Global Product Owner at Mondelēz International, about how global enterprises like Mondelēz are tackling the toughest challenges in operationalizing AI and GenAI at scale. From AI councils to legal frameworks, this conversation dives deep into the processes, pitfalls, and priorities behind production-ready AI. 🔥 Key Highlights: – Why 85% of GenAI POCs never reach production – The structure and role of the AI Council at Mondelēz – How to validate use cases across legal, compliance, and architecture – Techniques for minimizing hallucinations (beyond just RAG) – Real-world strategies for model monitoring and human-in-the-loop evaluation – How hybrid cloud infrastructure supports compliance in global AI projects – Balancing transparency and trust with measurable business value 🎙️ Expo Roundtable – Powered by Nscale: https://www.nscale.com/ – Produced by Data Phoenix: https://dataphoenix.info 🤖 AI INFRA SUMMIT 4 📍San Francisco, October 29 🎟️ Register now (35% Discount: DATAPHOENIX35): https://lu.ma/aiinfra_4?coupon=DATAPH... ⏱️Chapters 00:00 – Introduction 00:24 – Nitin's background and career path 01:00 – Current AI and GenAI projects at Mondelēz 01:50 – Why most GenAI POCs fail to go live 03:20 – Structuring AI governance and stakeholder input 04:48 – Addressing hallucinations with rules and feedback loops 06:02 – Evaluating answers and monitoring model quality 07:02 – AI compliance and hybrid infrastructure approaches 08:21 – Building trust and transparency with end users 09:00 – Closing thoughts 👉 If you enjoyed the discussion, be sure to like the video, share your favorite insights in the comments, and subscribe to stay updated on our latest talks and demos! 🔗 Follow us for more: – LinkedIn (Dmytro):   / spodarets   – LinkedIn (Data Phoenix):   / data-phoenix   – YouTube (Events):    / dataphoenixevents   – YouTube (Dmytro):    / @dmytrospodarets   – Twitter/X: https://x.com/Data_Phoenix – Telegram: https://t.me/DataPhoenix – Facebook:   / dataphoenix.info   – Website: https://dataphoenix.info/