How AI Transformed Coca-Cola's Global Supply Chain
Leading a 150+ person AI organization to build Transformer-based models, intelligent chatbots, and production MLOps infrastructure serving billions in global logistics.
The Challenge
Coca-Cola's global supply chain spans 200+ countries with complex logistics networks, fluctuating demand patterns, and massive operational data volumes. The organization needed to move from reactive decision-making to predictive, AI-driven operations — but lacked the infrastructure, models, and team to execute at enterprise scale.
Key challenges included siloed data across regions, no centralized ML platform, manual forecasting processes prone to error, and customer-facing operations that relied heavily on human agents for routine inquiries.
The Approach
Supply Chain AI
Designed and deployed Transformer-based models for demand forecasting and supply chain optimization. These models process signals across weather, promotions, regional events, and historical patterns to predict demand at granular levels.
Intelligent Chatbots
Built AI chatbots using AWS Lex, Dialogflow, and GPT-4, automating customer service and internal operations. Fine-tuned language models via HuggingFace and PyTorch for Coca-Cola's specific domain vocabulary and use cases.
MLOps at Scale
Stood up production MLOps infrastructure using MLflow, Weights & Biases, and Prometheus/Grafana for model monitoring. Built on AWS (EC2 GPU instances, EKS, Redshift) to support continuous training, deployment, and observability.
AI Partnerships
Applied AI to power partnerships with major entertainment and sports brands including the Olympics, NASCAR, and NCAA — creating personalized fan engagement, dynamic content, and predictive campaign optimization.
Tech Stack
The Impact
Under this leadership, Coca-Cola's AI organization grew to 150+ engineers and data scientists, transitioning from experimental notebooks to a fully production-grade ML platform. Demand forecasting accuracy improved significantly, chatbot automation reduced operational costs, and AI-powered partnerships generated measurable engagement lifts across global campaigns.
The MLOps infrastructure now supports continuous model training, automated deployment, and real-time monitoring — enabling Coca-Cola to iterate on AI solutions at the speed of business.
Facing similar supply chain or MLOps challenges?
I help enterprises build AI systems that actually make it to production — not just slide decks.
Book a Strategy Call