Case Study: Embark (self-driving truck company) accelerates safer autonomous trucking with Databricks

Ensuring long-haul truck safety with data and AI

Embark builds self-driving long-haul trucks to make highways safer and freight transport more efficient, but the project generates massive, high‑definition sensor datasets (video, LiDAR, RADAR) — roughly 37,000 hours — that were impossible to analyze at scale. Engineers were limited to 30‑second clips on in‑house laptops, faced slow and costly data transfers, and spent days of machine and engineering time just to validate model changes.

By running Databricks on AWS, Embark unified ingestion, ETL and model workflows, moved to hundreds of cloud nodes, and used interactive notebooks and Tableau dashboards for collaboration and monitoring. The platform unlocked over 35,000 hours of data, doubled the speed of neural‑network performance analysis, increased the on‑the‑fly analysis window by 75%, and cut offline evaluation from days to minutes — enabling much faster, safer iteration of their autonomous driving models.

Embark

Jason Snell

Lead Software Engineer