Buy tickets

CO-LOCATED EVENTS

CRYPTICA

DIGIHEALTH

PANDORA

AICOMMERCE

AICSI

GAID

Hotel Metropol, Belgrade

17th-21st November 2025

DSC Europe 25 SCHEDULE

Download Preliminary Schedule
NOV 17TH
VIRTUAL
NOV 18TH
VIRTUAL
NOV 19TH
LIVE
NOV 20TH
LIVE
NOV 21ST
LIVE
DATA & AI TALKS
BREAKOUT SESSIONS
CO-LOCATED EVENTS

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Ut elit tellus, luctus nec ullamcorper mattis, pulvinar dapibus leo.

MACHINE LEARNING
NATURAL LANGUAGE PROCESSING
COMPUTER VISION
Robotic Process Automation
Speech Recognition
Recommendation Systems
Predictive Analytics
Autonomous Systems
Expert Systems
GENERATIVE AI
Responsible AI
Data Engineering & Architecture
Emerging Tech

Ivo Andric B

Agentic AI,

ML Application

Registration

09:00

 –

09:30

Registration

Multimodal document processing at scale: How Lechler empowers its sales agents team with agentic AI

Agentic AI
Technical talk, Business talk
Intermediate
In this journey we want to show you how we at Lechler, Europe’s market leader in spray nozzles, built our own OrderAssistant within 6 months using the newest multimodal large language models and Microsoft Graph for a seamless Office 365 integration. We will show you which hurdles we faced, why off-the-shelf solutions did not work for us in a reality beyond shiny marketing slides, how we tackled them and how you could apply our learnings to your own buisness. Lechler is a medium-sized machine engineering company, processing more than 100,000 customer mails per year and offering a product portfolio with more than 40,000 products. Thanks to the help of our solution we could reduce the mundane work for our sales agents by thousands of hours per year, enabling them to focus on the really important topics: customer care and human to human interaction.
09:30

 –

10:00

Multimodal document processing at scale: How Lechler empowers its sales agents team with agentic AI

Michael Ikemann & Ozgur Yilmaz
Head of Data & AI @ Lechler GmbH & Associate Professor | Technical AI Consultant @ Adana Alparslan Türkeş Science and Technology University | Lechler GmbH

Intelligence Swarm Logic and Techno-Functional Multi-Agent Systems for Hyper-Accurate Knowledge Synthesis and Extraction

Agentic AI
Dive into advanced architectures where techno-functional teams and orchestrated AI agents collaborate seamlessly to automate, optimize, and scale information extraction workflows, achieving accuracy and adaptability beyond traditional approaches. This talk explores the inner workings of multi-agent orchestration, blending domain expertise and autonomous AI agents to build resilient, context-aware pipelines for complex data synthesis and extraction.
10:00

 –

10:30

Intelligence Swarm Logic and Techno-Functional Multi-Agent Systems for Hyper-Accurate Knowledge Synthesis and Extraction

Ivan Peric
Global Head of Artifical Intelligence R&D @ Synechron

From Prompt Engineering to Agentic AI: A Real-World Journey in Building Scalable LLM Applications

Agentic AI
AI Engineering has been evolving rapidly over the past few years and we see a shift from simple prompt engineering towards Context Engineering and Agentic AI. In this talk we will share real-world case studies that show the shift from static Retrieval Augmented Generation (RAG) workflows towards dynamic, multi-agent systems to solve complex challenges like evidence generation for bid writing. Teo, an AI Engineer at AutogenAI, will introduce the concept of agentic AI, highlight its practical implications, and showcase its evolution within Autogen’s product. She will highlight how her role has transitioned from traditional “Prompt Engineering” to “Context Engineering” – the art of providing all the necessary context for an LLM to plausibly solve a task – driven by the increasing sophistication of AI models. She’ll illustrate the limitations of the legacy “”search-and-give-results”” approach and the benefits gained from single and multi-agent frameworks, leading to more precise, adaptable, and user-centric features. Heiko, a GenAI Blackbelt at Google, will then provide a theoretical overview of AI agents and explain key agentic patterns and how they enable complex, iterative workflows. He will also perform a live demonstration of an agent-driven evidencing process, showcasing the dynamic, multi-step reasoning and collaboration between agents that addresses real-world user needs with higher accuracy and efficiency. Attendees will gain practical insights into designing, implementing, and deploying sophisticated LLM-powered applications using AI agents, understanding how to move beyond the limitations of traditional prompt engineering to build scalable and robust AI solutions.
10:30

 –

11:00

From Prompt Engineering to Agentic AI: A Real-World Journey in Building Scalable LLM Applications

Heiko Hotz & Teodora Danilovic
Generative AI Global Blackbelt @ Google & AI Engineer @ AutogenAI

Agentic AI: What businesses can’t ignore

Agentic AI
Demystify the anatomy of Agentic AI and dive deeper into how this formidable technology is going to transform the businesses forever.
11:00

 –

11:30

Agentic AI: What businesses can’t ignore

Astitva Karunesh
AI Business Lead @ Lloyds Banking Group

Coffee Break

11:30

 –

12:00

Coffee Break

Keynote

Keynote
TBA
12:00

 –

12:45

Keynote

TBA
TBA

Keynote

Keynote
TBA
12:45

 –

13:30

Keynote

TBA
TBA

Lunch Break

13:30

 –

14:30

Lunch Break

TBA

ML Application
TBA
14:30

 –

15:00

TBA

Nikolay Burlutskiy
Senior Manager GenAI Platforms @ Mars

TBA

ML Application
TBA
15:00

 –

15:30

TBA

Oday Almajed
General Manager – Data & AI @ Government Expenditure & Projects Efficiency Authority

Coffee Break

15:30

 –

16:00

Coffee Break

TBA

ML Application
TBA
16:00

 –

16:30

TBA

Marco Bettiolo
CTO @ Sensay

TBA

ML Application
TBA
16:30

 –

17:00

TBA

TBA
TBA

Day Recap

17:00

 –

17:15

Day Recap

Ivo Andric A

Data Engineering,

Augmented Automation

Registration

09:00

 –

09:30

Registration

Automatic DWH Migration – On-prem to Cloud and More

Data Engineering
Nowadays, Business Intelligence is a mature technology. Technological migrations are more frequent than one might imagine. Regardless if we are moving to cloud or just reengineering old integration jobs, rewriting ETL from scratch in a new technology might be looming over our heads. As one might imagine, doing this manually might be an absolutely colossal task. Hundreds if not thousands mutually interdependent jobs. Written by somebody else years ago, often badly. Knowing what the job is actually supposed to do is more often an exception than a rule. For this purpose, we can use tools for automatic ETL migration from technology to technology. They offer obvious benefits, but also have limitations. In this lecture, we will dive deeply into the World of ETL migration. When to do it automatically, when to do it by hand? What are the threats on our journey? Last but not the least, we shall provide some real world experiences and examples. Success stories, perhaps an epic failure or two. There is only one way to find out – attend the DSC Europe 2025 Conference!
09:30

 –

10:00

Automatic DWH Migration – On-prem to Cloud and More

Miljenko Vukovic
Data Engineer, Founder @ VIS

Architecture problems of streaming AI pipelines using public social posts sentiment analysis as an example

Data Engineering
In many cases, social posts containing anger or depression precede crimes committed by younger people in their education institutions, such as schools, colleges and universities. Sentiment analysis of such posts could be a good basis to track rolling risk scores for students. While developing a prototype of such a system, I faced typical streaming processing issues which I would like to discuss in this talk. Traffic spikes and inference jitter: sometimes when high risk is detected, the system must react immediately, so we have to guarantee predictable latency. External dependencies and quotas: social media APIs and AI instruments impose limits and fail intermittently, but the system still needs to react quickly. Event delivery and consistency (“exactly once”): it is important not to raise risk score twice for the same action. Time and ordering: late and out-of-order events distort windowed aggregations and dashboards. State & checkpoints: how do we track what has already been processed? Schema and contract evolution: most certainly the system will evolve, we must design for multi-version interoperability. Performance issues related to agent2agent message exchange: the sheer volume of data that needs to be serialized and deserialized may be a problem. Alert quality & observability — shifting FP/FN rates, model/data drift, and weak end-to-end tracing make root-cause analysis hard. We’ll look into these issues and review the pros and cons of the typical solutions and their combinations, such as different levels of caching, using self-hosted solutions instead of cloud ones, backpressure and throttling mechanisms, smart retries and the like.
10:00

 –

10:30

Architecture problems of streaming AI pipelines using public social posts sentiment analysis as an example

Vadim Opolski
AI Engineering Guild Lead & Senior Data Engineer @ DXC

Reproducible ML training datasets from code: JetBrains’ approach

Data Engineering
Technical talk
Beginner to Intermediate
JetBrains builds model-ready datasets from code to train Mellum, our purpose-built code-completion LLM. We’ll walk through the pipeline: corpus assembly, license detection and filtering, language/file normalization, multi-level deduplication, de-contamination to avoid eval leakage, and secrets/PII scanning. We’ll show how these processes feed versioned Iceberg snapshots and dataset cards used to train Mellum (a 4B-parameter model with an 8k context window, trained on ~4.2T tokens) and share what worked—and didn’t—in achieving strong IDE acceptance rates. You’ll leave with a practical checklist for turning public code into trustworthy, reproducible training data.
10:30

 –

11:00

Reproducible ML training datasets from code: JetBrains’ approach

Sergei Boitsov
Data Engineer @ Jetbrains

TBA

Data Engineering
TBA
11:00

 –

11:30

TBA

TBA
TBA

Coffee Break

11:30

 –

12:00

Coffee Break

AI is Only as Good as Your Data Engineering

Panel
A panel that highlights how foundational data engineering work impacts the success (or failure) of machine learning initiatives — from pipeline reliability to real-time data access and governance.
12:00

 –

12:45

AI is Only as Good as Your Data Engineering

Panel

Black

12:45

 –

13:30

Black

Lunch Break

13:30

 –

14:30

Lunch Break

Beyond the Bot

Augmented Automation
Technical talk, Research talk
Intermediate to Advanced
Imagine AI that works together, not alone. This talk shows you the exciting world of collaborative AI agent clusters – think of it as an ‘Agent Orchestra’ in action! We’ll explore how to orchestrate these intelligent teams using the adaptable Semantic Kernel, Azure AI Foundry, and the Azure AI (Cognitive) Services. Discover the architectural blueprints that enable multiple AI agents to operate autonomously yet harmoniously, tackling complex challenges that would overwhelm a solo AI. You’ll gain actionable insights into seamlessly integrating these intelligent networks into your existing applications, unlocking new AI-powered potential without a complete system rebuild. Join me to learn how to conduct your own AI symphonies!
14:30

 –

15:00

Beyond the Bot

Miodrag Cekikj
Founder @ Adopt Intelligence

AI-Powered Process Intelligence: The Future of Automation

Augmented Automation
Business talk
Intermediate
The future of automation is not about replacing people, it’s about empowering organizations with AI-driven process intelligence. This session reveals how detailed business analysis fuels smarter, adaptive automations that go beyond efficiency to create real business transformation. With practical examples, we’ll show how to connect process knowledge and AI technology to unlock agility, scalability, and measurable results.
15:00

 –

15:30

AI-Powered Process Intelligence: The Future of Automation

Marko Pantic
Head of Automation and Process Department @ Simplify Oursourcing d.o.o.

Coffee Break

15:30

 –

16:00

Coffee Break

From Ledgers to Transformers: How AI & Cloud Are Rebuilding Financial Systems

Augmented Automation
Technical talk, Business talk, Research talk
Intermediate to Advanced
Finance may be the last frontier of AI, but it’s quickly being redefined. In this talk, I’ll share lessons from building production-ready AI and cloud systems for financial operations like automating billing pipelines, ERP/CRM integrations, and real-time data flows that power revenue recognition and reporting. I’ll also preview insights from my PhD research on applying transformer-based NLP models to financial time series forecasting. Attendees will learn how to design scalable, cloud-native data pipelines (using Go, Python, and Google Cloud Run), where automation and AI deliver the most impact in finance, and what the future of decision intelligence looks like when accountants and data scientists converge. This session bridges technical architecture with applied AI, showing how we can transform one of the most traditional domains into a data-driven backbone for business.
16:00

 –

16:30

From Ledgers to Transformers: How AI & Cloud Are Rebuilding Financial Systems

Joseph Marks
Lead Engineer – Financial Systems & Software Engineering; Co-Founder, Subledger @ Madhive.com. , Subledger.app

TBA

Augmented Automation
TBA
16:30

 –

17:00

TBA

TBA
TBA

Day Recap

17:00

 –

17:15

Day Recap

TESLA B – DSC:X

Data in Cloud,

Applied AI

Registration

09:00

 –

09:30

Registration

Lost in performance

Data in Cloud
Use case, Transformational talk
Intermediate to Advanced
Join us for a critical examination of how modern cloud infrastructure has inadvertently created a culture of performance complacency, costing organizations millions in unnecessary spending. In this presentation, we’ll explore how the convenience of cloud scaling has led engineering teams to abandon performance optimization in favor of simply adding more resources. This “throw hardware at the problem” mentality is driving up infrastructure costs by 30-40% annually while creating over-provisioned systems that operate at less than 30% efficiency. What You’ll Learn: – Why cloud bill shock is becoming the #1 barrier to digital transformation, – How performance optimization skills have eroded in the cloud-first era, – Real-world cost impact: companies spending 3-5x more than necessary on infrastructure, – Practical FinOps strategies and cloud-native tools to regain control, – How performance discipline creates sustainable competitive advantages. Key Takeaway: Organizations that master cloud performance optimization don’t just save money—they build faster, more reliable systems while their competitors struggle with bloated infrastructure costs.
09:30

 –

10:00

Lost in performance

Boris Perkovic
CEO & Solutions Architect @ Scaletech Platforms

TBA

Data in Cloud
TBA
10:00

 –

10:30

TBA

TBA
TBA

TBA

Data in Cloud
TBA
10:30

 –

11:00

TBA

TBA
TBA

Coffee Break

11:00

 –

11:30

Coffee Break

Black

11:30

 –

11:47

Black

Keynote

Keynote
TBA
11:30

 –

12:15

Keynote

TBA
TBA

Keynote

Keynote
TBA
12:15

 –

13:00

Keynote

TBA
TBA

Lunch Break

13:00

 –

14:30

Lunch Break

Black

13:30

 –

13:43

Black

TBA

Applied AI
TBA
14:30

 –

15:00

TBA

TBA
TBA

AI Readiness in Practice: Building the Foundations for Adoption — with AI Itself

Applied AI
At DSC Europe 25, Rakan will present a practical framework for AI readiness — highlighting the essential data, governance, and platform foundations required for sustainable AI adoption. He will also explore how AI itself can be used to accelerate the development of these foundations, while avoiding common pitfalls such as rushed deployments, technical debt, and short-sighted wins that undermine long-term value.
15:00

 –

15:30

AI Readiness in Practice: Building the Foundations for Adoption — with AI Itself

Rakan Aalsoraye
Director – Data, AI & Advanced Analytics @ Tawuniya Insurance

Coffee Break

15:30

 –

16:00

Coffee Break

Grounding your AI Strategy in business value to avoid the Pilot Trap™

Applied AI
Transformational talk
Intermediate
“With MIT’s State of AI 2025 report highlighting that 95% of organisations fail to realise value from their initial AI investments (yet), and many projects stuck in the Pilot Trap™, your approach to AI adoption is pivotal. We all see the qualitative value in AI’s ability to transform our organisations. Turning that opportunity into reality is the key business challenge of our time and one I’ve worked with hundreds of executives at large enterprises to understand and help them overcome. This talk outlines a framework by which your organisation can position, plan, and execute an AI strategy that builds your organisational AI capabilities and empowers you to turn opportunity into success. You’ll hear about important factors from your organisational strategy to be applied to AI, how to assess your current capabilities and gaps, how to align AI to your organisation’s unique opportunities and differentiators, and a pragmatic approach to getting started. “
16:00

 –

16:30

Grounding your AI Strategy in business value to avoid the Pilot Trap™

Matt Quinn
TBA

AI meets documents: From chatbots to AI-powered editing

Applied AI
Discover how AI is transforming the way we work with office files, from simple chatbot interactions to advanced AI-powered editing tools. We’ll explore integration options for AI and documents as well as showcase innovative features of ONLYOFFICE, including a smart AI agent. Whether you’re an AI enthusiast, a productivity seeker, or a tech-savvy professional, this session will provide actionable insights into leveraging AI for smarter document management and editing
16:30

 –

17:00

AI meets documents: From chatbots to AI-powered editing

Petar Zivanov
Sales Manager @ Only Office

Day Recap

17:00

 –

17:15

Day Recap

* this program is not final and is subject to change, full schedule will be available soon

Download
Billetterie Weezevent