iFood AI Logistics Optimization — 12% Distance Reduction and 50% Idle Time Reduction

iFood's AI-driven logistics optimization cut delivery distance by 12%, reduced courier idle time by 50%, and lifted delivery SLA from 80% to 95% across more than 39 million monthly orders in Brazil, Mexico, and Colombia. "The use of such algorithms sped up the process of discovering new parameters and, thus, our Logistics area is far more efficient today. We deliver more today than two years ago," Sandor Caetano, iFood's Chief Data Scientist, said in an AWS case study. Before the rollout, iFood — founded in 2011 as a Latin American foodtech leader operating alongside 220,000 restaurants and 170,000 registered couriers in more than 1,000 cities — had AI initiatives scattered across business units but no centralized platform. In 2019 the company committed $20 million in shareholder capital to create the Artificial Intelligence Academy, a dedicated organization focused on machine learning, deep learning, and logistical efficiency. The logistics team needed an infrastructure that could simulate routes at scale and put validated algorithms into production quickly, with traffic only spiking at peak times. iFood standardized on Amazon Web Services — Amazon SageMaker for model training and serving, Amazon Elastic Kubernetes Service for inference workloads, Amazon Kinesis for streaming data, and Amazon DynamoDB for low-latency lookups — and built an internal route simulator that analyzes multiple operating parameters per weekday and time band. Models run in a safe AWS sandbox first, and only proven ones get promoted into the live logistics stack; the same pattern powers the recommendation models that surface restaurants and dishes tailored to each customer's taste. The deployment feeds directly into iFood's delivery-time estimates shown to customers in the app. Looking ahead, iFood plans to expand the Academy's scope into horizontal use cases such as computer-vision image enhancement for restaurant menus and ingredient extraction from dish photos.