Spark sessions 008

The Spark Sessions are a recurring event of the Data Science Initiative at the Faculty of Computer and Information Science. They are designed to facilitate the exchange of ideas and inspiration. The event consists of several snappy (5-7 min) presentations, followed by a casual gathering with refreshments.

When? May 19 2026 at 18:00.

Where? Faculty of Computer and Information Science, Večna pot 113, Ljubljana.

What?

Using Telematics and Machine Learning to Promote Eco-Driving
by Tina Šteblaj (Senior Data Analyst, Zavarovalnica Triglav)

Insurance company Triglav collects driving telematics data through its mobile application Drajv. To provide customers with clear feedback on the eco-friendliness of their driving behavior, a data-driven smoothness score was developed based on telematics signals. The talk will demonstrate how telematics can be used to quantify driving smoothness and how customers can be differentiated according to distinct driving styles, with the aim of promoting safer and more sustainable driving.

Homography: From Pixel Space to the Real World
by Blaž Pridgar (Data Scientist, Valira AI)

Sometimes the most useful tools are the simplest. Homography maps points from one plane to another using nothing more than a 3×3 matrix. This talk covers the intuition and math behind it, and where it shines. Whether this is new to you or a refresher, the goal is the same: to show how textbook linear algebra can be surprisingly powerful in a modern AI pipeline. We’ll close with a production example where we used homography to generate real-world measurements from detected wheel positions.

Not all FHIR MCPs are created equal
by Matic Bernik (Data Scientist, Better)

Modern healthcare systems store data across multiple interoperability standards and services, making reliable AI access much harder than simply connecting an LLM to a database.

This talk focuses on a FHIR agent built around an MCP (Model Context Protocol) server - one component of Better’s broader “Conversational EHR” initiative, a conversational layer composed of multiple specialized healthcare agents.

Spec-Driven Development with AI Agents
by Martin Jurkovič (Data Scientist, Viaduct AI)

A major source of inconsistency, unreliability, and code bloat in AI-assisted development is the use of unstructured prompts. This talk presents a spec-driven workflow, where work is divided into specification, implementation, and validation phases. By introducing versioned artifacts, task-level isolation, test driven development, and cross-agent reviews, the approach improves reliability, traceability, and alignment between intent and implementation.

Want to attend? Attendance is free. Apply here.

Interested in giving a talk at one of the future events? Apply here!