PyData Trójmiasto #44

44th edition of PyData Trójmiasto coming up!

Where: Łużycka 8B, Gdynia, Tensor Z building.
When: 24th of June at 6PM

Registration: please register with your full first and last name. Don't forget to bring your ID when entering the event venue.
RSVP available until 6pm day before. In any urgent communication please let us know at kontakt@pydata-trojmiasto.pl

Parking: Free parking located along the street next to the building.

Agenda:

18:00 - 18:05 - Event boarding
18:05 - 18:10 - A few words about PyData
18:10 - 18:55 - From Annotation to Deployment:
Building an Object Detection Pipeline with Geti, YOLO26, and OpenVINO by Adrian Boguszewski
19:00 - 19:45 - Prompt, Pray, Pay: The New Economics of AI Coding Assistants by Marcin Duszyński
19:45 - Networking

From Annotation to Deployment: Building an Object Detection Pipeline with Geti, YOLO26, and OpenVINO by Adrian Boguszewski & Francesco Mattioli

Learn from Ultralytics and Intel® AI experts working side by side in this hands-on session and discover how to build production-ready computer vision pipelines for real-world manufacturing and industrial scenarios. We’ll guide you through a complete, end-to-end object detection workflow—from dataset creation and model training to optimization and edge deployment.

Start by rapidly annotating and iteratively refining image labels with Geti™, enabling efficient dataset bootstrapping even with limited data. Using this dataset, we’ll fine-tune an Ultralytics YOLO26 model for custom object detection—using hazelnuts as a practical example of common manufacturing inspection tasks. The trained model is then converted and optimized with OpenVINO™, leveraging techniques such as hardware acceleration and precision optimization to achieve low-latency inference. Finally, we’ll deploy and benchmark the solution on Intel® Core™ Processors Series 2 and Intel® Core™ Ultra processors, highlighting key considerations for latency, throughput, and resource utilization in edge environments.

Developers will leave with actionable insights to build industrial-grade AI pipelines:

End-to-End Solution: From image annotation to edge deployment, making vision AI practical for real-world applications.
Industrial Relevance: Apply object detection to manufacturing and inspection tasks to boost automation and efficiency.
Rapid Development: Accelerate dataset creation using Geti™, reducing time-to-deployment.
Optimized Edge AI: Deploy high-performance models with OpenVINO on Intel® hardware for low-latency, efficient inference.

Prompt, Pray, Pay: The New Economics of AI Coding Assistants by Marcin Duszyński

How token billing, context size, loaded toolsets, and model choice change how we use AI coding assistants, and how to control these costs.

AI coding assistants used to feel like unlimited magic behind a monthly subscription. As pricing shifts toward usage and tokens, we need a new mental model.

This talk breaks down what really drives the cost: context size, tool overhead, model choice, and so on. Then we look at how to control it: trim context, pick the right model for the task, and a few other practical moves.

The goal is not to use AI less. It is to use it deliberately, with better quality and fewer billing surprises.