Conversational Analytics for Cyber Physical Systems: AI Framework for Mixed Protocol Log Intelligence

This work presents an AI conversational analytics framework that unifies heterogeneous cyber physical system logs, spanning CAN traffic, Ethernet frames,
telematics, calibration data, and IT security events, into a single natural language investigative environment. Supporting diverse formats such as log,
pcap, csv, parquet, and mf4, and powered by high performance processing libraries, the system enables scalable, cross protocol correlation at fleet and
enterprise scale.

Beyond describing the architecture, the talk will walk through how the system actually works under the hood: how generative AI and NLP
components are embedded, how mixed telemetry is indexed and queried, and how anomalies and multi layer relationships are analyzed.

Attendees will also learn how to replicate these capabilities within their own engineering toolchains, including practical patterns for integrating LLMs, building unified layers, and operationalizing conversational analytics inside existing environments. 

Ivan Granero

Senior Security Engineer, Bosch 

Ivan Granero is an Automotive Cybersecurity Expert and has 15 years of development experience in Automotive Engineering (Passenger Cars, Commercial Vehicles – Fleets) in areas of acquisition, design, validation and manufacturing. He coordinates Bosch’s Americas commercial vehicle cross-divisional product security activities. He holds an MS in Software Engineering with a concentration in Cybersecurity from the University of Texas at El Paso and a BS EE, enjoys participating in Capture-the-Flag competitions and recently competed at DEFCON31 and 32 as part of BOSCH/ETAS team in the Car Hacking Village CTF obtaining 3rd place.

Ivan holds a MS in Cybersecurity from the University of Texas at El Paso. He is a PhD candidate of Engineering in Artificial Intelligence and Machine Learning (2027).