AI Meetup Frankfurt April 2026

Agenda
18:00 – 18:10; Reception
18:10 – 18:15; Welcome note from organiser & host/sponsor
18:15 - 18:45; Speaker I: Dr. Cory Whitney
Title: Forecasting Under Uncertainty: Using LLMs and Monte Carlo Simulation to Generate a Multiverse of Intervention Models
Abstract: When planning complex interventions, we usually build one model, but that model hides assumptions and blind spots. This talk presents a framework that uses LLMs to rapidly generate multiple plausible stakeholder models, then runs Monte Carlo simulations across all of them.
The key insight: disagreement is the signal. Convergence across simulations builds confidence; a subset predicting failure is a red flag. The question shifts from "is our model right?" to "under what conditions does this intervention fail?" This reframes AI as a generator of structured uncertainty. The simulations turn ambiguity into a measurable, actionable signal.

18:45 - 19:15; Speaker II: Prasad Tilloo - Remote on Zoom
Title: Building with AI Without Being an AI-Expert: What Vibe Coding Actually Looks Like
Abstract: I'm a Solutions Architect, not an ML engineer — but I recently used vibe coding and modern AI tools to build two working AI projects from scratch. In this talk, I’ll walk you through the real process: how I used Claude, Cursor, and Gemini to accelerate development, what the AI got right, what broke, and what I had to figure out myself to make things actually work. The goal isn’t a product demo — it’s an honest look at what’s possible (and what’s not) when a non-AI specialist tries to build with these tools today.

19:15 - 19:45; Break & Networking

19:45 - 20:15; Speaker III: Fabian-Robert Stöter
Title: AI Is Not Only Text and Images: A Case Study in Audio-AI
Abstract: LLMs and GenAI get most of the attention today. But audio has been using deep learning in production for about 50 years — before computer vision, and long before language. The field has its own data, benchmarks, evaluation challenges, and modeling paradigms. This talk gives a primer on audio-AI for a modern deep learning audience.
We first look at the main tasks in audio-AI for speech and music, and unpack how sound is represented, trained, and measured. Along the way, we compare these models to models from computer vision and explain why audio is so hard. We also cover the open-source tools and datasets that shape the field. From there, we dive deeper into 3 topics: Music source separation getting single tracks — vocals, drums, bass — out of a full mixture; Speech separation pulling apart two or more people speaking at the same time; Permutation-invariant training and what makes real conversations difficult. Automatic lyric transcription recognizes song text, including punctuation, capital letters, and song structure.
At the end, we turn to the open questions that remain across all three areas.

20:15 - 20:30; Announcements
20:30 - Socializing

Join Zoom Meeting - AI Meetup Frankfurt - April 2026
https://us02web.zoom.us/j/86222400498?pwd=ixyuHbxXD7GOzbRcHZSSKdL7bkauBQ.1

Meeting ID: 862 2240 0498
Passcode: 181724
---

Interested in speaking?
Link to our Speaker Form: Google Form

Interested in sponsoring?
Partner with us to support the AI community and gain visibility among industry professionals and enthusiasts.
Link to our Sponsorship Form: Google Form