XConf Europe 2024

Track 1

11:00 

Building Data Mesh on Databricks and Azure for a large manufacturing company

Rieke Heinze and Dominika Makuch
 

It is hard to find detailed technical recommendations for implementing Data Mesh. Thoughtworks is currently supporting a large European manufacturing company on their Data Mesh journey using a concrete technology stack: Databricks on Azure. In this talk you will learn about concrete examples of libraries and services we used to implement Data Mesh principles. We will also share lessons and obstacles we faced, as well as things we found missing. For example, did you know - how great the support is that Unity Catalog gives you when it comes to securing your data? - that the brand-new Databricks Asset Bundles are awesome in supporting CI/CD? - that you can implement computational governance on the infrastructure side with Open Policy Agent and that you can do it using TDD and trunk-based development?

11:55 

Lesson learned implementing a large-scale Data Mesh transformation at Payback

Alessandro Confetti and Norbert Wirth
 

Data Mesh aims at solving the challenges of modern data-driven and data-intensive organizations with 4 key pillars: domain-driven data ownership, data as a product, self-serve data platform, federated computational governance. Translating these pillars into a sound tech roadmap and organizational transformation is a journey where every decision or step matters for better or worst. In this talk we will share the lessons learned from a the three-year Data Mesh journey at PAYBACK as where to start from and what to start with (e.g. tech, data, teams, domains, organization); how to build full data ownership without generating unnecessary overhead on people; how to prevent data product fragmentation; what to look for when migrating data and processes from data warehouses and data lakes into data products; and, most importantly, how to make sure the entire organization benefits from these initiatives.
 

12.50 

Unlocking Backstage’s potential: insights and solutions for a smoother Developer Experience

Germelinda Musliaj and Abel Guillen

In the dynamic landscape of software development where efficiency and collaboration are paramount, Backstage stands out. Like any tool, realizing its full potential demands more than just setting it up - it calls for experimenting and tinkering around to find what capabilities work better in your context. In this talk, we’ll take on a journey with Backstage, sharing practical tips for solving common problems you might run into. From customizing it to fit your team's needs, through improving onboarding experience, removing undifferentiated work to automated governance we’ll go over real experiences and lessons learned to showcase how to make the most out of its capabilities.

Track 2

11:00 

How to measure your software's impact on the planet

Minette Mangahas and Andy Symonds

The Green Software Foundation’s Impact Framework aims to make the environmental impacts of software easier to calculate and share. The Impact Framework allows you to calculate the environmental impacts, such as carbon, of your software applications without writing any code. This talk will introduce the Impact Framework using some fun storytelling (a panda and racoon in space!). The project is entirely open source and composability is a core design principle. Will will do a live demonstration of how you can develop your own plugins and plug them into the framework, or pick from a broad universe of open source plugins created by others. We want this talk to inspire you to start using the Impact Framework as you develop software, by making it as simple as possible.

11:55 

Cloud Cost Chaos? Get Clarity with FinOps

Maria Valero

When having applications in the Cloud, it is easy to end up with unmanaged cloud costs and unclear developer ownership. I am a FinOps practitioner and have experienced lots of war stories of developers battling cost chaos without direction. FinOps appears as a practical approach for bringing clarity and collaboration to cloud cost management.
In this session, you will get practical recommendations to start your FinOps journey and develop the capabilities that you will need and how to implement a company-wide FinOps practice.
You should be able to walk away with actionable steps and resources to unleash the power of FinOps in your cloud journey.

12.50 

Uncovering seams in the Mainframe & GenAI-assisted reverse engineering for incremental Legacy modernisation

Tom Coggrave and Alessio Ferri

Mainframe systems handle much of the world's computing workload, but adding new features to them can be challenging due to their complex architecture. This complexity also makes them difficult to understand and replace. To mitigate these risks, an incremental approach to legacy displacement is recommended, gradually replacing legacy components with modern technologies. This strategy involves introducing seams into the mainframe system—points where logic flow can be diverted to newer services. In a recent project, we explored various methods to create these seams in a long-lived mainframe system and will share our findings and how they relate to other legacy modernization approaches. Join us to discuss incremental legacy modernization and learn about CodeConcise, a Proof of Concept developed in-house by EMPC and Data and AI service lines. 

14:30

Scaling Websockets in Charge Station Management System

Prachi Tyagi

Electric Vehicle (EV) adoption is rapidly increasing, leading to a surge in the demand for efficient Charging Station Management Systems (CSMS) - a central nervous system, orchestrating charging operations across networks of charging stations. The Open ChargePoint Protocol (OCPP) facilitates communication between charging stations and CSMS via web socket connections, enabling bi-directional message exchange. However, the stateful nature of web socket connections, compounded by OCPP, presents unique challenges in scaling CSMS. In this talk, we explore various approaches to address these challenges. Join us as we discuss practical strategies for building robust and scalable EV charging infrastructure.

15:25

What digitalization for a critical, highly-distributed infrastructure company means

Andreas Reisenauer (from 50Hertz)
 

Transmission System Operators (TSOs) play an instrumental role in enabling the transition to renewable energy. The distributed & critical nature of energy infrastructure places unique challenges for technology and the overall ‘digitalization’ of the energy landscape. Modernizing the whole software and infrastructure stack while bridging gaps between operational technology (OT) and information technology (IT) poses a cross-functional challenge. In this talk. Andreas will demonstrate the technological approach to delivering a platform that enables a cloud-native environment for realizing data-intensive applications at scale that will drive core TSO processes. From building infrastructure in data centers, to enabling seamless delivery and providing building blocks of a modern data architecture is part of this effort.

16:20

Container Security - Zero trust & principle of least privilege in containers inside K8s

Amit Dube
 

As organisations are rapidly moving to containerised workload, the security design inside the containers has become very critical. As part of this talk, - What are various security options available inside k8s API definition for pod and how they work in conjunction and complement each other? - I will explain the security issues related to k8s default permission set when nothing is explicitly defined (more than 90% of containers use this). - How to start from zero permissions and add what we need based on the application in the container. - What are possible options to enforce/audit it in a big platform (landing zone) inside an organisation using admission controllers and policy engines. How to scale it? - How to transition from no security to zero trust & principle of least privilege model in bigger organisations?

14:30

Inclusive employee experience: AccessibilityOps for Compliance & Inclusion

Sara Michelazzo

Building accessible internal software is no longer optional. Dive into AccessibilityOps, our approach to digital accessibility compliance. TechOps, Thoughtworks' IT department, shares real-world experience and working software. We'll showcase how we build and monitor accessibility across products, engage with vendors to ensure compliance, and build awareness in our delivery teams. Learn how to integrate accessibility into your development lifecycle and product strategy. Why now? Prioritizing accessibility empowers employees with disabilities, fosters inclusivity, and ensures compliance with regulations like the upcoming European Accessibility Act impacting private companies by June 2025.

15:25

Evaluating LLM systems: the nuts and bolts

Aili Asikainen and Oege Dijk

Recent developments in large language models have made it easier than ever to develop AI systems. Companies are building internal and external chatbots, RAG systems and copilots. Many of these projects run into a similar roadblock: it is very quick and easy to get to the first 80% useability, but really hard to cover the remaining 20% to actually deliver a successful product. In this talk we will argue that the key to delivering a successful AI product is systematic evaluation of your system so that you can continuously and confidently improve. We will show practical examples ranging from how to evaluate LLM outputs and agent based systems to understanding new usage patterns in production. These insights enable fast iteration and informed decision making that drive both technical improvements and product development.

16:20

Fine-tuning small GenAI models for straight forward and  great results

Meissane Chami and Siobhan Hughes

As organisations are rapidly moving to containerised workload, the security design inside the containers has become very critical. As part of this talk, - What are various security options available inside k8s API definition for pod and how they work in conjunction and complement each other? - I will explain the security issues related to k8s default permission set when nothing is explicitly defined (more than 90% of containers use this). - How to start from zero permissions and add what we need based on the application in the container. - What are possible options to enforce/audit it in a big platform (landing zone) inside an organisation using admission controllers and policy engines. How to scale it? - How to transition from no security to zero trust & principle of least privilege model in bigger organisations?

Rebecca Parsons, CTO Emerita

Magdalena Kuhn (BMW) and Anton Vlasov

Andrew Harmel-Law, Technical Principal