Preferences Over Benchmarks: Model Routing — Archana Kamath & Tyler Gillam, DigitalOcean
AI Engineer World's Fair 2026 · 15:53
Cloud and AI infrastructure
DigitalOcean provides cloud computing and AI infrastructure for developers and businesses. Founded in 2012, it began with Droplet virtual machines and expanded into managed Kubernetes, databases, storage and application hosting.
Its inference services offer a catalog of models with serverless and dedicated deployment options. Teams can use DigitalOcean-hosted models and third-party commercial models through the platform.
Inference Router selects models for individual requests using task requirements and preferences such as cost and latency. Teams can configure routing policies and model pools, then evaluate how the resulting choices perform on their workloads.
AI Engineer World's Fair 2026 · 15:53
Affiliations reflect their AIE appearances, not necessarily current employment.