Stop Treating LLMs Like REST APIs

Programming Stage — 45 minutes

Why do PoCs run smoothly while launch day implodes? Because LLM traffic is a streaming, state-heavy beast that breaks every REST assumption: requests aren’t stateless, payloads snowball with context, and GPU memory melts under token floods. We’ll map the three checkpoints where most projects stall—context explosion, batch backfires, cache chaos—and show how LLM-D’s open-source sharding plus a hybrid NVIDIA/AMD node pool turns each choke point into a green light. You’ll see live before-and-after dashboards, get a YAML ladder you can drop into any cluster, and learn a back-of-the-napkin formula to keep cost per 1 000 tokens under control. Read More...

• Jeff Fan [Senior Solutions Architect — DigitalOcean]

Driving Impact with Email Marketing: Insights from a Data-Driven Campaign Project"

Programming Stage — 30 minutes

In this session, Mihalci Ioan, together with Adelina Moldovan and Radu-Corneliu Diaconu, will highlight how targeted communication and performance analytics can elevate customer engagement. A brief overview of the broader role of AI in marketing will also be shared, offering a high-level perspective on how emerging technologies are shaping campaign planning and execution.  Read More...

• Ioan Mihalci [Operational Management Director — MassMutual Romania]
• Adelina Moldovan [Project Manager/ Product Owner — MassMutual Romania]
• Radu Diaconu [Senior Business Systems Analyst — MassMutual Romania]

How to Build AI-Native Tools to Enhance Your Productivity

Programming Stage — 45 minutes

This talk explores how developers can build their own assistants, pair programmers, code reviewers, and more by teaching how to build AI agents and multi-agent systems (MAS) on demand to meet the needs of any developer for their productivity.  A great example of this is ChatGPT's Agent Mode: how does it work? How can we build it ourselves? This talk will shed light on this topic and more, enabling developers to create their own intelligent assistants. Read More...

• Tejas Kumar [AI Developer Advocate — IBM]

Keynote powered by msg

Programming Stage — 30 minutes

Lunch Break

Programming Stage — 1 hour

The future & challenges of cloud

Programming Stage — 45 minutes

This opinionated talk will delve into the insights and innovations that are shaping the future of Cloud Computing. It will cover a range of topics including cloud architectures, data gravity, multi-cloud strategies, security, compliance, and the future of AI & Cloud.  The session will highlight emerging trends such as AI-native cloud platforms, serverless 2.0, and composable cloud architectures, while also addressing the challenges of multi-cloud workloads, vendor lock-in, and balancing performance with cost efficiency. It will discuss the impact of data gravity on cloud strategies and the importance of standardization. Security and compliance will be another focal point, with insights into AI-powered cyberattacks, zero trust adoption, and the role of Cloud Security Posture Management (CSPM) tools. Finally, it will look ahead to the future of AI & Cloud, examining hardware innovations, the move to higher-level services, and the rise of no-code, low-code, and AI-assisted coding platforms. This session promises to provide valuable knowledge for anyone looking to navigate the rapidly changing landscape of cloud computing

• Anders Lybecker [Principal Architect — Microsoft]

Turbocharging Aftersales - The Role of Cloud & AI in the Evolution of Vehicle Diagnostic

Programming Stage — 30 minutes

Let’s start with a simple question: What truly defines a vehicle today? Is it the engine, the design, or perhaps the driving experience? From the perspective of modern automotive development, the answer is increasingly clear—software. Vehicles are no longer just machines; they’re complex, software-defined systems on wheels.But with great software comes great complexity. Diagnosing issues in these vehicles is no longer about just popping the hood and checking for mechanical issues. Today, we’re dealing with millions of lines of code, dozens of electronic control units, and cutting-edge technologies that demand a whole new level of expertise. These challenges extend well beyond diagnostics and into the realm of aftersales. While complexity is rising, customers still expect timely updates, efficient repairs, and satisfactory customer experiences. How do we, as developers, create software solutions that can simplify these complexities and support technicians and service teams in the field?In this talk, we’ll explore how cloud-native technologies and artificial intelligence are helping us tackle these challenges head-on. From smarter diagnostics to streamlined aftersales processes, we’ll look at how our work as developers is shaping the future of automotive aftersales. Read More...

• Otto Fischer [Software Engineer & Team lead, Automotive Industry — MHP – A Porsche Company]
• Paul Ciesla [Software Architect, Automotive Industry — MHP – A Porsche Company]

The Unwritten Rules of Fast React Apps

Programming Stage — 45 minutes

Performance in React isn’t just about tools, it’s about patterns, habits, and choices that often go unnoticed. In this talk, we’ll uncover the “unwritten rules” behind fast React apps, the ones that aren’t in the docs but show up in real-world teams, codebases, and bugs. You’ll learn practical strategies to avoid unnecessary re-renders, manage state more efficiently, and structure components for performance by default. We’ll cover smart rendering boundaries, bundling tips, when (and when not) to memoize, and how to debug slow components without losing your mind.Whether you’re working on a side project or a large-scale production app, this talk will help you build React apps that don’t just work, they feel fast. No magic, no hype, just patterns that actually work. Read More...

• Hammad Hassan [Software Engineer — Independent]

AI-powered Development: From Concept to Prototype in Minutes with Claude Code

Programming Stage — 30 minutes

Product teams have ideas. Engineering teams have backlogs. The time between 'what if we built this' and seeing it working often kills good ideas before they get proper evaluation.  Watch as we build micro-frontend applications from scratch using Claude Code in a couple of minutes, showcasing automatic integration with the Design System, proper event-driven communication patterns, and adherence to architectural best practices. Interested in learning more? Join us to discover how!  Read More...

• Andrei Eftimie [Principal Developer — 8x8]
• Géza-Tamás Kovács [Senior Engineering Manager — 8x8]

A New Hope in the Cloud

Programming Stage — 30 minutes

In this galaxy, cloud infrastructure is dominated by a few hyperscalers, some with an increasing history of GDPR breaches and opaque data practices. But there’s a rebellion forming. This session introduces our approach at a new OpenStack-based platform in its early phase, built for everyone who wants an open, transparent, and community-driven alternative. You’ll hear how our team’s upstream contributions strengthen OpenStack, why open source is the key to trust, and how building in the open gives businesses and developers freedom from lock-in and misuse of data.The empire doesn’t have to strike back—let’s build a different future. Read More...

• Radu Nicoara [Senior Software Engineer — EPAM Romania]

Java's Unfair Advantage: Why LLMs Write Better Enterprise Code in Java

Java Stage — 45 minutes

Java's radical openness created the perfect training ground for LLMs: open specifications, open source implementations, and open standards. LLMs didn't just read Java code - they absorbed the entire Java Community Process, every JSR, every JEP, the complete MicroProfile and Jakarta EE specifications. They understand not just syntax, but the reasoning behind every architectural decision, every design pattern, every enterprise best practice.

Through live demonstrations, I will show you how LLMs and agents can be used to develop enterprise-grade code that adheres to standards by default and is intended for both LLMs and humans.
This code can maintain consistency across entire codebases and follow custom rules.
Feel free to ask questions at any time. There's nothing wrong with interactive keynotes! Read More...

• Adam Bien [Java Enthusiast — Independent]

Where are Java and Kotlin in the AI world?

Java Stage — 30 minutes

Is Java keeping pace with the fast-evolving AI technologies that seem to be popping up everywhere? Let’s take a look at the efforts being made within the Java ecosystem to stay relevant in this era.  We’ll explore high-level frameworks like Spring AI, which offers seamless integration with popular AI service providers, as well as low-level advancements such as the Vector API and CUDA support. If you love Java but feel hesitant about diving into AI, this talk is for you.  Read More...

• Silviu Cucuiet [Software Engineer — Cognizant Romania]

Code Quality in the AI era, needed more than ever

Java Stage — 45 minutes

As developers, we participate every day in our software life cycle adding new logic, adapting the existing one, and integrating with services or platforms.
But, do we put the same effort into making software robust, maintainable, consistent, secure, clear, and tested?Moreover, when developers are adding code generated by an AI code generator, can we rely on that code?In this session, I will show the need for good practices to have a great Code Quality along with the issues of not using it, using free and open-source tools.Code Quality goes beyond the produced software and its programming language. Is a fundamental part of a developer's role growth, something that represents us no matter the language and the company where we work. Do not lose the thread. Read More...

• Jonathan Vila [Senior Developer Advocate — Sonar]

Keynote powered by Google

Java Stage — 30 minutes

Lunch Break

Java Stage — 1 hour

High-Performance Spring Persistence

Java Stage — 45 minutes

In this presentation, we will go through the various performance bottleneck issues that are prevalent in Spring Data JPA applications. Not only will we identify the core problem, but I’m going to show you the most effective solutions to solve these issues.

This is a hands-on talk, as we will go through code examples that I will share with the audience as well.

If you are a Java developer who has writes data access code using Spring, JPA, and Hibernate, then you are going to benefit from these high-performance tips. Read More...

• Vlad Mihalcea [Java Champion — Hypersistence]

Efortless event filtering from heavy Java logic to Flink queries

Java Stage — 30 minutes

The process of triggering a notification is usually based on a single explicit request that comes from the client application. Have you ever thought of triggering notifications via your application? At ING, we use Flink SQL to convert heavy Java logic into streamlined queries using SQL-like syntax. This allows us to perform complex data transformations. During the live-coding session, you will see how easy it is to filter Kafka events with the help of Apache Flink. We will transition from a Java based application to SQL like queries to demonstrate development time and performance savings. Using Flink in our demo, we will consume events from an input topic, filter them based on specific rules and publish the resulting events on an output Kafka topic.  Read More...

• Andreea Tugui [Software engineer — ING Hubs Romania]
• Alin Pandichi [IT Area Lead — ING Hubs Romania]

Keynote powered by Deloitte Technology Center

Java Stage — 45 minutes

Partner Keynote

Java Stage — 30 minutes

Roundtable session 2

Roundtables — 1 hour

Roundtable session 3

Roundtables — 1 hour