AI Austria Meetup #4 - Innsbruck

AI Austria is back in Innsbruck for the 4th edition of our local community meetup series! This time we are sponsored by the Wirtschaftskammer Tirol.

This meetup is part of our growing AI community in Tyrol, with a clear technical focus. We go beyond high-level overviews and look into how models, systems, and real projects actually work.
The main goal is to learn from each other and to connect industry and academia. Expect technical talks, practical insights, and space for real discussions with people who actively build and research AI.
This event is fully in English, so internationals are very welcome :)

Agenda
🕡 6:00 - 06:30 pm // Open Doors, Arrival
🕡 6:30 pm // Start of Event, Welcome

🕡 6:45 pm // 🎙 Francesca Finotello, LFU

From mixed expression signals to cell types: an overview of in silico deconvolution

RNA sequencing (RNA-seq) provides high-dimensional measurements of gene expression from complex tissues, yielding a digital profile of all active genes in a sample. However, these measurements reflect mixtures of multiple cell types, creating a fundamental computational challenge: how can we recover the underlying cellular composition from aggregated signals?

In this talk, I will present computational deconvolution methods, which model bulk RNA-seq data as mixtures of cell-type-specific signals and infer the abundance of distinct cell populations. I will outline the key algorithmic ideas behind these approaches, from early regression-based models to more recent methods that integrate single-cell references and leverage advances in machine learning. I will discuss key challenges arising from biological and technological variability, and how modern methods address them.

I will further highlight recent extensions that move beyond bulk measurements to enable spatially resolved inference of tissue organization. Finally, I will discuss how these approaches can be applied in cancer research to derive clinically relevant biomarkers, improving patient stratification and prediction of therapeutic response.

🕡 7:30 pm // 🎙 Leon Kogler, CASABLANCA hotelsoftware

From AI Research to Industrial Reality: Rethinking Automated Testing

At CASABLANCA hotelsoftware, we started with an ambitious goal in the GENIUS ITEA4 research project: using AI to automate REST API test generation. This resulted in RESTifAI, a system that outperformed several state-of-the-art approaches and led to a paper accepted at ICSE 2026.
During this work, we realized that existing evaluation metrics barely reflect the real effectiveness of generated REST API tests. To address this, we developed RESTestBench, a benchmark for evaluating REST API tests generated from natural language requirements, accepted at EASE 2026.
But the most important lesson came afterwards: even with strong research results, industrial adoption of REST API tests remained difficult due to false positives, complex dependencies, microservices, mocking, and the effort required to maintain realistic testing environments. We realized we had started with one of the hardest testing strategies while still lacking strong fundamentals such as sufficient unit test coverage.

This talk is about the gap between research and industry, why adapting direction can sometimes be more valuable than strictly following the original plan, and what we learned from shifting our focus from highly ambitious automation approaches toward more practical and established testing strategies.

🕡 8:15 pm // Food, Drinks and Networking - We are thinking of something to help new people, who might come to the event alone and don't know anyone, get a little bit more included! :)
🕡 9:15 pm // end

Location: Wirtschaftskammer Tirol, Wilhelm Greil Straße 7, Innsbruck

Date & Time: 17.06.2026, 6:00 – 9:15 pm
Entrance: free
Sponsors: WKO

✨ Everyone’s welcome — get ready for an evening of learning from eachother, great people, and inspiring discussions!

Let’s shape the future of AI together — right here in Innsbruck! 🤖💡
We are looking forward to meeting you!