PyData Trójmiasto #33 (Hapag-Lloyd venue)

We are welcoming you to join 33rd edition of PyData Trójmiasto meetup!

This time the event will be held at Hapag-Lloyd Knowledge Center venue - we're very thankful for such an opportunity and recommend you to join us as well!

• When: 26th of March, at 18:00 (6PM)
• Where: Hapag-Lloyd Knowledge Center, Al. Grunwaldzka 413
• Registration info - important!

Number of seats is limited to 50. Please provide your full first and last name and email address while registering for the event, here on meetup. In case of any urgent changes please leave us a note via [email protected] . Don't forget to bring your ID for the security check.

About "Building RAG Systems for Enterprises"

Operationalizing Retrieval-Augmented Generation (RAG) systems at scale presents unique challenges, including the need for seamless integration, customization, and adaptability to diverse organizational requirements. Let's discuss lessons we've learned from deploying these solutions, explore essential tools every RAG developer should have at their disposal, and identify common pitfalls to avoid. I'll also explain why we've decided to open-source ragbits—our collection of foundational building blocks for GenAI applications.

Mateusz Hordyński - bio

I'm a software engineer who specializes in creating big data architectures for both cloud and on-premises setups. Currently, I'm a Technical Leader at deepsense.ai, where I design generative AI applications and build data pipelines to support them. I'm also the lead maintainer of the open-source project db-ally. Outside of work, I try to live the digital nomad lifestyle, which has taught me how to work remotely from some pretty weird office setups.

About "Unlocking Efficiency: AI-Driven Email Classification for Salesforce Customer Support"

As the shipping and logistics industry continues to adopt artificial intelligence to enhance operational efficiency, companies like Hapag-Lloyd are exploring innovative applications of AI to automate tasks, optimize logistics, and improve customer support. This presentation begins with a brief introduction to the AI initiatives that Hapag-Lloyd is engaged in, followed by a case study on one of the key areas of focus for our company which is the efficient classification of the high volume of emails we receive daily. This classification task is crucial for effective customer support in Salesforce (SF) but our current setup struggles with accurate classification of emails, leading to performance problems. Our analysis of internal data reveals significant imbalances in classes distribution, performance differences across areas and languages, and the need for improved data preprocessing. The limitations of Salesforce's built-in classification models, including the lack of preprocessing, limited training case selection, and blackbox models, led us to explore the development of a custom model. We will present our adventure towards development of an improved custom AI-driven email classification model, including the creation of a ground truth dataset and leveraging BERT. Our results show promising improvements in classification accuracy, and we will discuss our plans for integrating our custom model with Salesforce.

Michał Papaj - bio

Michał Papaj is a Data Scientist at Hapag-Lloyd. His current focus is on vessel schedule maintenance, although he has been involved in a range of AI projects over the past two years. Prior to joining Hapag-Lloyd, Michal worked at Intel, where he contributed to the development of speech-to-text solutions. He holds a Ph.D. in the area of Digital Signal Processing.

Marek Blok - bio

Marek Blok is a Data Scientist at Hapag-Lloyd, who for almost two years has been developing AI solutions for berth load prediction and customer email classification. Prior to joining Hapag-Lloyd, Marek worked at the Gdańsk University of Technology, where he completed a PhD and DSc in Digital Signal Processing, with a focus on standard and ML-aided digital signal processing and analysis of telecommunication signals.

See you at PyData!