Discover Brainly's tech stack, designed to future-proof JavaScript architectures in preparation for AI-driven assistant development. This keynote explores how Gene 🧬 framework streamlines development workflows, enhances scalability, and sets the stage for AI integration. Learn about its modular design and systematic practices that enable rapid adaptation and robust system evolution. This session offers a deep dive into how cutting-edge JavaScript practices can prepare your projects for the next technology cycle. Join us to see how Brainly prepared JS tech stack for the AI future.
Curious about whether a micro-task or a macro-task comes first? Wondering how to manage long-running tasks to prevent your browser from freezing? Or how to wait until the browser paints the next frame? If these questions leave you scratching your head, join us as we demystify the JavaScript event loop. You’ll learn how the browser schedules tasks and discover practical insights into managing task execution. We’ll also explore intriguing APIs like requestIdleCallback and take a peek into the future with the upcoming Scheduler API. No frameworks or libraries here—just pure, classic JavaScript.
This presentation focuses on the fascinating world of generative music, with a primary focus on JavaScript implementation. During the presentation, we will move into a world of controlled chaos and use code to create dynamic, unpredictable musical compositions.
How to optimize the loading speed of web applications while ensuring Security and Interactivity? During this presentation, I will demonstrate why Streaming and Partial Prerendering will change the way web applications are created. I will show the direction in which web applications are heading and discuss whether frontend developers will soon need to become fullstack developers.
Music production ain't easy, just like asynchronous programming! Add RxJS to the mix and we are on the rocket science level. Historically we had callbacks… Promises are the standard now but who likes standards in JS world? We need something new and complex! Something REACTIVE! Lets take a look at RxJS and how to smoothly transition from promises to observables and tap out some rhythm in the process.
Next.js may be the go-to for many, but have you ever checked out Nuxt? It’s not just a Vue.js counterpart - it might be a game-changer for you. During this talk, we’ll break down why Nuxt offers a superior developer experience with its smooth workflows, built-in features, and robust ecosystem that can save you time and … money. Whether you’re building complex apps or just want a framework that feels intuitive, Nuxt delivers the flexibility and ease that can really surprise you. Let’s dive in and see why it could be the perfect alternative to take your projects to the "NEXT" level!
Large language models (LLMs) are not just tools for supporting programming or answering everyday questions. They can become a key element of your backend, working with existing IT infrastructure. Are such solutions always efficient, safe and economical? Unfortunately, no. Fortunately, there are ways to effectively use the potential of LLMs. I invite you to the lecture, especially if you approach the topic of AI with extreme skepticism.
Is it wise to run code from strangers? Well, we do it all the time and there's no backing out of it. Let's take a look at how a JavaScript project can proactively defend itself from supply chain attacks, assuming it already ships with a malicious dependency. Limit access to globals for each package? Sure. Control if a package can access network or file system? Yup, that too. And no more prototype pollution. I will show how future features of JavaScript being discussed in TC39 can be used to protect your project right now and even execute actual malware live for your entertainment.
Build your own GenAI laboratory for images and video! Leverage the latest models trending on your feeds and learn how to run them independently! Create, experiment, and amaze (even yourself)! During this session, I’ll show you how to get started with ComfyUI, what additional tools to install to give your work a real boost, and what to focus on to achieve results as quickly as possible. Face swapping? Specific poses? Videos? Lip-syncing? Relighting? Style transfers from other images? It’s all within your reach - simpler than you might think! A blend of theory and practice will let us explore the philosophy of working with ComfyUI, the types of elements (nodes and plugins) used in real-world and everything you need to know to extend your lab with Computer Vision, LLMs and more.
In this presentation, we will delve into the intricate process of training large language models, using Bielik 2.0 as a case study. We will examine the critical aspects of Dataset and Data Collection, Model Architecture, Training Process, Evaluation and Performance Metrics, and Future Directions in the field. By focusing on these elements, we aim to provide a comprehensive understanding of how large language models like Bielik 2.0 are developed and optimized.
The politics and mythology of the new technology present the development of AI as an inevitable, determined and total process although for now only the investors are determined. From this perspective, large models are modern pyramids, which are to demonstrate power and provide immortality for the pharaohs. But what if other mythologies and other technologies are possible? What about small forest models, or models that, hidden under the stairs, guard the house, feeding on sun, water or, why not, leftovers from dinner? In my talk, I will show several turning points in the emergence of artificial intelligence technologies and myths, and I will also try to outline the possibilities of creating your own small, autonomous models.
We spent over 6 months implementing our internal AI assistant. We learned a lot in the process and in this talk we want to share our most important findings, learnings and problems we encountered along with our solutions to them. You will learn what to avoid and what to focus on when planning one for yourself, or your company.
In this session, we'll explore how prompt engineering introduces non-deterministic elements into deterministic coding environments. We'll discuss strategies to manage and control unpredictable outputs, ensuring more reliable and consistent results when working with AI-driven systems.
LLMs are bad at jokes. They just... won’t get it. But why? How about instead of learning how LLMs behave in certain ways, you learn how they work? Once you grasp it, their behavior will be no surprise. No joke! I will tell you about: transformer-based architecture, what causes hallucinations, what is fine-tuning and when it makes sense, the science behind prompt engineering, when to use AI agents, AI architectural patterns for complex problem-solving.
Getting your data ready for RAG systems can be tricky, but it doesn't have to be a headache. In this session, I'll share some practical tips from my projects, on how to make this process smoother. We'll cover various methods of breaking your text into chunks for better LLM context. I'll show you how I format documents for better results and how to handle references between paragraphs, documents, and images. We'll also tackle how I process more complex documents, like legal texts or websites to make those more machine-friendly, without losing important details. Lastly, I'll give you insights on tweaking human-created content to make it work better for machines. There will be memes
The world of programming is facing new challenges. For years, No-Code and Low-Code solutions have been gaining popularity, undermining the unique value of the work of skilled programmers. Additionally, artificial intelligence, represented by tools such as ChatGPT, Claude 3.5 Sonnet or Cursor, is getting better at generating code every week. In this talk, we will analyse the impact of these technologies on the programming profession, with a particular focus on web developers. We will outline potential AI development scenarios and their implications for the IT industry. We will discuss whether programmers will maintain their position or be replaced by autonomous AI systems in the coming years.
In this talk, I will give an introduction to quantum machine learning. First, I will explain the basic concepts and algorithms of quantum computing, how it differs from the classical way of computing, what can be the possible applications of quantum computing, and what are the biggest challenges in this field. Then, I will present the theory behind quantum machine learning algorithms and talk about their possible applications and state of the art. Finally, I will present some quantum computing and quantum machine learning frameworks and libraries, and show how to get started in this field and where to learn more.
meat lunch
vege lunch
meat lunch
vege lunch