Can an LLM understand your car? Real telemetry and predictive insights

09:50–10:30 CET · Speakeasy · Track: AI

Modern cars produce far more data than what appears on the dashboard: sensor readings, diagnostic codes, pressure changes, temperature values, service intervals, and component health trends.

In this talk, we’ll explore how to turn telemetry collected from different real Porsche sports cars into clear, useful insights using a structured LLM-based analysis pipeline. We’ll detect measurable signals such as declining health, warning states, critical sensors, performance patterns, and upcoming maintenance needs. Then we’ll use an LLM as an explanation layer to summarize driving behavior, anomalies, and maintenance signals.

The goal is not to let the LLM control the car or replace expert inspection, but to make complex telemetry easier to understand, explain, and act on.

Speaker

• Soumaya Erradi, Senior Software Developer @ Atlantis

“Can an LLM understand your car? Real telemetry and predictive insights” at DevFest.cz 2026, 30 October 2026, Uhelný Mlýn, Libčice nad Vltavou, Czech Republic.