Data Sanity Talks

✨ Spotlight on the next poster ✨ Senior ML Engineer Milan Bradonjić will present a poster “Leveraging Geometric Graph Structures for Efficient Sparse Attention in LLMs”. We explore Spatial Graph Transformers, which replace standard quadratic self-attention with sparse attention patterns derived from geometric graph structures. In this approach, we investigate how geometric relationships in embedding space capture syntactic and semantic dependencies. Our goal is to establish new directions for efficient Transformers by bridging graph theory, geometric deep learning, and NLP. 🗓 October 4 | 12:00 — 22:00 📍 Startit Centar Belgrade ➡️ Event Agenda & Tickets: datasanity.dev

🔥 Next up in our lineup of poster presenters: Co-Founder at UkisAI, Jovan Kis, will present a poster titled “Bridging the Gap: Making AI Coding Tools Better at Building AI-Centric Products.” Jovan has created a tool that turns general AI coding assistants into powerful, up-to-date development platforms capable of building full-stack, multi-agent RAG chatbots with a UI — all from a single prompt. 🗓 October 4 | 12:00 — 22:00 📍 Startit Centar Belgrade ➡️ Event Agenda & Tickets: datasanity.dev

In a world where hype balloons keep inflating, we aim to bring the weight of real AI expertise. At Data Sanity Talks Belgrade on October 4, you’ll meet experts from leading tech companies who will share real industry insights — rather than sell cloud AI castles. Plus: connect with professionals, find like-minded enthusiasts, and enjoy free coffee, snacks, and the occasional random beer! 🎟 Get a discount with code DATASANITYOCT4 🗓 October 4 | 12:00 — 22:00 📍 Startit Centar, Belgrade ➡️ Full agenda & tickets: datasanity.dev

✨ Meet our next poster: “On-Device Intelligence for Poultry Farmers: Disease Detection with ML”. Nikola Mekić, Mobile Developer at XG Dynamics, will present his app that uses Computer Vision to automate the detection of poultry diseases on farms, significantly reducing the cost of health monitoring. 🗓 October 4 | 12:00 — 22:00 📍 Startit Centar Belgrade ➡️ Event Agenda & Tickets: datasanity.dev

🎙 Another talk coming up: “Architecture Problems and Solutions in Streaming AI Pipelines: A Case Study” by Vadim Opolski — Senior Data Engineer @ DXC From big data to real-time alerts: Vadim will share a streaming architecture for monitoring text, photos, and video with Spark and Kafka, highlighting key pitfalls and solutions. 🗓 October 4 | 12:00 — 22:00 📍 Startit Centar Belgrade ➡️ Event Agenda & Tickets: datasanity.dev

🔥 Meet our first poster presenter 🔥 Mariia Fedorova, Doctoral Research Fellow @ University of Oslo in her poster "Open Multilingual Web Datasets for LLM Training" Mariia will tell us about the HPLT project that is designed for LLM pretraining and versatile NLP tasks like translation, classification, and text generation. It provides large-scale multilingual datasets and pretrained models for ~200 languages, including Serbian, Croatian, and Bosnian. 🗓 October 4 | 12:00 — 22:00 📍 Startit Centar Belgrade ➡️ Event Agenda & Tickets: datasanity.dev

✨ Next in the discussion section: Marina Mijatović, CEO and Co-Founder of AngryLynx, leads the discussion “AI in the Workplace: Acceleration or Unrealistic Expectations?”. Together we’ll explore whether AI truly drives productivity and innovation, or whether hype sets us up for disappointment. A business-focused discussion on how AI reshapes work performance: does it truly make tasks easier and faster, or does it push employees under new pressure as managers expect more in less time? 🗓 October 4 | 12:00 — 22:00 📍 Startit Centar Belgrade ➡️ Event Agenda & Tickets: datasanity.dev

✨ Meet our next presenter ✨ Dušan D. Nešić, Founder @ AI Safety Serbia Dušan will host an open-table discussion “Why AI Attunement is Important: AI as the Other” — a fireside-style conversation unpacking the limits of alignment, the challenges of defining human goals, and how “attunement” could offer a more human-centered path for AI. 🗓 October 4 | 12:00 — 22:00 📍 Startit Centar Belgrade ➡️ Event Agenda & Tickets: datasanity.dev

✨ AI Assistants for Emotional Support: Good or Evil? Our first discussion announcement! Are AI tools a shortcut to affordable mental healthcare and more effective healing, or a way to cheat at being human? Join psychoanalyst Elizaveta Musatova for an open-table discussion to explore where human psyche meets artificial empathy. 🗓 October 4 | 12:00 — 22:00 📍 Startit Centar Belgrade ➡️ Event Agenda & Tickets: datasanity.dev

🔥 Up next on the mic: Olga Titova, AI Product Manager @ Wargaming In her talk “When Updates Feel Like Breakups: Designing Safe and Responsible AI Companions” Olga explores how AI companions are shaping human relationships, the psychological risks of judging them by engagement metrics alone, and how we can design them to support healthier, more meaningful lives. 🗓 October 4 | 12:00 — 22:00 📍 Startit Centar Belgrade ➡️ Event Agenda & Tickets: datasanity.dev

✨ Spotlight’s on… Ana Stojković Knežević, Project Manager @ Auxality AI is moving faster than management and regulations — can leaders keep up? In her talk “Business in the Age of AI: Trends, Management, and Safety” Ana will show how business and data science leaders can drive innovation responsibly, keeping people's trust. 🗓 October 4 | 12:00 — 22:00 📍 Startit Centar Belgrade ➡️ Event Agenda & Tickets: datasanity.dev

✨ Another awesome speaker ✨ Stanislav Tsepa — ML Engineer @ FAANG, flying in from the UK ✈️ In his talk “When LLMs Train Your Models: Using AI to Build Better AI” Stanislav will show how AI can actually help create smarter AI — from cleaning noisy datasets and prompt-optimization to generating synthetic data and distilling knowledge into smaller models. Stanislav will also give us practical examples of using LLM-judge, automatic label generation, and active learning with LLMs to boost real-world pipelines. 🎟 Don’t miss out: 25% off tickets with code DATASANITYFLASH25, valid until the day ends. 🗓 October 4 | 12:00 — 22:00 📍 Startit Centar Belgrade ➡️ Event Agenda & Tickets: datasanity.dev

💡 Next on the list 💡 Aleksandar Anđelković — Principal Engineering Manager @ Microsoft In his talk “A Practical Guide to LLM Context Engineering”, Alexandar introduces a practical framework to overcome LLM context limits by selecting, compressing, and structuring information with RAG, summarization, memory, and agentic workflows to build smarter, more reliable AI applications. ✨️️️️️ Your exclusive 25% off is waiting — unlock it with code DATASANITYFLASH25, today only. 🗓 October 4 | 12:00 — 22:00 📍 Startit Centar Belgrade ➡️ Event Agenda & Tickets: datasanity.dev

🤖💀 To AI, or not to AI: that is the question. Whether ’tis wiser in the mind to embrace The promises and perils of this novel tech, Or to resist the tide of change, And by resisting, be left behind? Explore this very question — not in verse, but through talks, discussions, posters, and networking with experts from Microsoft, Meta, JetBrains, Foursquare and more! 🎟 Secure your spot now with 25% off tickets, code DATASANITYFLASH25 (valid until Sept 25). 🗓 October 4 | 12:00 — 22:00 📍 Startit Centar Belgrade ➡️ Event Agenda & Tickets: datasanity.dev

🔹 Spotlight on our next speaker: Dušan Zamurović 🔹 Engineering Manager at Foursquare, Dušan will talk about “Building a Data Platform as a Single Source of Truth”. This talk shares a real-world journey of building an AWS-based, event-driven data processing platform that matured from backend infrastructure into a company-wide self-service hub. It highlights key technical steps, cultural shifts, and DS/DE collaboration that built trust and empowered business teams to explore and act on data. 🗓 October 4 | 12:00 — 22:00 📍 Belgrade, Startit Centar 🎟 Info & Tickets: https://datasanity.dev/

Message from our media partner Belgrade Hub Coworking Space ✨ Collaborate, network, and share ideas in an inclusive and supportive environment in Belgrade Hub Coworking Space! With the perfect blend of productivity and comfort, we are nestled in the charming neighborhood of Banovo brdo, Belgrade. 🔗 https://www.belgradehub.rs/ 📍 Požeška 83a, Belgrade

Message from our media partner CodeRetreat ✨ CodeRetreat — лагерь для взрослых в Черногории. Хайкаем, перезагружаемся и учимся посреди невероятной красоты природы в душевной компании интересных людей! 24 - 27 октября проводим смену для Data-инженеров! Умеем готовить идеальный баланс отдыха и пользы: уедете не только с новыми знаниями, но и новыми друзьями и тележкой впечатлений. 🔗 Все подробности: https://coderetreat.me/datacamp2 🎟️ Используй промокод "datasanity", чтобы получить скидку

🌟 Meet our next speaker: Mariya Zlenko 🌟 AI Product Director at Uralsib, joining us from Azerbaijan ✈️ In her talk, “Launching AI in Conservative Corporations: Turning Skepticism into Value”, Mariya will share her experience of bringing AI to a place where people are sceptical about the new technology, and what she's learned along the way. 🗓 October 4 | 12:00 — 22:00 📍 Belgrade, Startit Centar 🎟 Info & Tickets: https://datasanity.dev/

Here is a video from our recent event: To AI or Not to AI? Integration & Adoption Case Study Paul Katurov dives into: 🟢 How is AI being integrated into different processes and departments within a company? 🟢 What challenges might you face? 🟢 When is it better to rely on humans? +++++++++++ You can get more practical insights and personally ask AI experts about your specific topic at the upcoming Data Sanity Talks Belgrade on October 4. 🚀 Info & Tickets: datasanity.dev

🚀 Our next speaker for Data Sanity Talks! 🚀 Elena Ericheva — AI scientist at JetBrains, flying in from Montenegro ✈️ Her talk: "Mesa Optimisation in Large Models and AI Safety" Elena will explain what happens when big AI systems start behaving in ways we didn’t directly plan for — and why that matters for keeping them safe to use. 🗓 October 4 | 12:00 — 22:00 📍 Belgrade, Startit Centar 🎟 Info & Tickets: https://datasanity.dev/