Automotive Technology | Autonomous Vehicles | AI
Building the AI brain for autonomous vehicles, on a startup budget, at fleet scale
Halo, a UK autonomous vehicle innovator, needed a cloud architecture capable of ingesting terabytes of real-world fleet data daily, training vehicle-aware AI models, and doing it all cost-effectively on a startup budget. Presidio designed and deployed an AWS-based platform that handles the full data lifecycle, ingest, categorize, train, delete, and hosts Halo’s V2X platform for global fleet deployment.
THE OUTCOME
Halo, a UK autonomous vehicle innovator, needed a cloud architecture capable of ingesting terabytes of real-world fleet data daily, training vehicle-aware AI models, and doing it all cost-effectively on a startup budget. Presidio designed and deployed an AWS-based platform that handles the full data lifecycle, ingest, categorize, train, delete, and hosts Halo’s V2X platform for global fleet deployment.
HOW THEY GOT THERE
Presidio worked closely with Halo to map data ingestion requirements across a growing multi-vehicle fleet, then designed a tailored AWS architecture using S3, IoT Core, Lambda, SageMaker, Glue, and ECR. Data is stored only for the duration of training, then automatically and securely deleted. The Cloud Management Portal adds real-time visibility into compute, storage, and pipeline performance, with granular billing transparency.
WHY IT WORKED
Halo’s AI approach incorporates surrounding infrastructure, other vehicles, and human behavior, not just the vehicle itself. Training those models requires diverse, real-world fleet data at scale. The architecture Presidio built gives Halo the throughput to train continuously on live data while keeping cloud costs lean enough for a scaling startup to sustain.
“The new cloud architecture that Presidio delivered drives our mission to accelerate the development of autonomous vehicle technology in the UK.”