Sept 22 - Stuttgart Physical AI Workshop and Meetup

Join us at Impact Hub Stuttgart on September 23rd for the Stuttgart Physical AI Workshop and Meetup, co-presented by Nebius and Voxel51.
Seats are limited, Pre-registration is mandatory.

Date, Time and Location
Sep 22, 2026
5:30 PM - 8:30 PM CEST
Impact Hub Stuttgart, Quellenstraße 7a, 70376 Stuttgart

Meetup Speakers

A Multi-Modal Pipeline for Automated Building Energy Performance Estimation at Urban Scale

We present an automated pipeline for estimating building-level heating demand from heterogeneous urban data sources, without requiring on-site audits.

The system extracts the physical building parameters required by the DIN 18599 monthly heat balance standard from three complementary sources: street-level imagery processed through a facade analysis pipeline, cadastral records from the German real estate register, and three-dimensional building models.

The aggregated features are passed to machine learning models that predict the physical inputs to a deterministic DIN 18599 calculation, producing EPBD-compliant heating demand estimates at urban scale.

The target applications include municipal heat planning under Germany's Heat Planning Act (Wärmeplanungsgesetz), renovation potential screening in support of EPBD Minimum Energy Performance Standards, and ESG-driven portfolio risk assessment for financial institutions operating under EU Taxonomy reporting requirements.

By removing the dependency on costly on-site audits, this project makes systematic building energy assessment viable at the resolution and pace that current climate and regulatory mandates demand.

Speaker: Dhia Abdelli is a Data Scientist at Lumanaa, specializing in data strategy, data modeling, and building scalable data platforms. He applies machine learning and AI to a range of real-world use cases, including manufacturing, to turn data into practical solutions and actionable insights.

Meetup Agenda

5:30–6:30 PM

• Networking, food, drinks, and lightning talks

Workshop Agenda

This hands-on session uses DROID, a real-world robotics dataset loaded into FiftyOne as a native multimodal MCAP recording, and YOLO11n, fine-tuned live during the session.

6:30–7:30 PM

• Welcome + framing: from raw robot logs to a trained detector
• Explore a real DROID robotics recording in FiftyOne's native multimodal MCAP viewer: camera, proprioception, and language on one synced timeline, no ROS install required
• Curate: extract and browse frames from the recording, filter and deduplicate
• Compute embeddings on the curated frames; explore via similarity search and embeddings visualization (via Nebius Serverless AI Jobs)

7:30–8:00 PM

• Auto-label: open-vocabulary detection to generate bounding boxes for the robot gripper and target objects
• Train: fine-tune a YOLO11n detector on the auto-labeled frames (via Nebius Serverless AI Jobs)
• Evaluate results and close the loop: view predictions back on the original MCAP timeline

8:00–8:30 PM

• What else Nebius offers: Token Factory walkthrough — chat/vision models, fine-tuning, credits