CO-LOCATED EVENTS

Hotel Metropol, Belgrade

18th-22nd November 2024

DSC Europe 24 SCHEDULE

Ivo Andric B

Computer Vision & Robotics

 

Registration

09:00

 –

09:30

Registration

AI in maritime – Autonomous ships

Computer Vision & Robotics
Technical talk
Advanced
AI has strongly entered the maritime industry, and although it is perceived as a traditional industry, the impact and application of AI is increasing. It is used in different ways and for various purposes, for example as an aid in predicting ship maintenance, for determining the optimal route, navigation and ship tracking, ship management in canals, for manipulation and logistics in ports, and for ensuring safe navigation and for an autonomous ship. Artificial intelligence solutions for autonomous navigation that we developed for automatic recognition of the state of the sea according to the Beaufort scale and detection of objects in maritime navigation such as ships, boats, but also small objects such as rafts, buoys, various fishing marks, swimmers and the like.

The developed models are based on state-of-the-art deep neural networks and visual transformers and can automatically recognize the state of the sea and detect objects at sea in real time and under realistic conditions from sea images taken during navigation in different world seas. The research was carried out as part of the EU project INNO2MARE.
09:30

 –

10:00

AI in maritime – Autonomous ships

Marina Ivasic Kos
Dean @ Faculty of Informatics and Digital Technologies, University of Rijeka and Centre for Artificial Intelligence, University of Rijeka

Building the Future – Unpacking the Essential Components of High-Performance and Sustainable AI Clusters

Computer Vision & Robotics
Technical talk, Business talk
Advanced
This presentation examines the essential components of high-performance AI computing infrastructure and its sustainable future. The discussion covers the dramatic evolution of AI computing capabilities, from the Titan supercomputer of 2012 to the GB200 systems of 2024, highlighting key improvements in performance and efficiency. The presentation introduces AMBER’s vision for an AI factory of the future through its aiterra “Sustainable Metal Cloud” initiative, which aims to establish a sustainable AI computing infrastructure using immersion cooling technology in Germany/Europe.
10:00

 –

10:30

Building the Future – Unpacking the Essential Components of High-Performance and Sustainable AI Clusters

Thomas Kitzler
Sales Director, EMEA @ Amber

Multimodal video face liveness — a better alternative to active liveness?

Computer Vision & Robotics
Technical talk
Advanced
This talk explores multimodal video face liveness as a more robust alternative to traditional face liveness detection techniques. We examine the limitations of single-frame passive and video-based active methods, highlighting failure cases. By combining depth estimation, motion analysis, and temporal data, multimodal video liveness offers a more comprehensive approach. We’ll discuss the use of foundational models, the integration of multiple modalities like RGB and depth, and how leveraging these inputs improves accuracy and scalability in facial authentication systems. Examples of real-world applications and datasets will be provided.
10:30

 –

11:00

Multimodal video face liveness — a better alternative to active liveness?

Efim Boieru
Manager of Machine Learning Engineering @ incode

From Banks to Smart Cities: How Biometric Technologies Are Shaping Tomorrow’s Landscape

Computer Vision & Robotics
11:00

 –

11:30

From Banks to Smart Cities: How Biometric Technologies Are Shaping Tomorrow’s Landscape

Dmitry Markov
CEO @ VisionLabs

Coffee Break

11:30

 –

12:00

Coffee Break

Keynote: The tech giants in AI – who is going to win the battle?

Keynote
12:00

 –

12:45

Keynote: The tech giants in AI – who is going to win the battle?

Ido Engel
Microsoft Cloud & AI GTM lead, SouthEast Europe @ Microsoft

Keynote: Key lessons from building Data and AI systems over the last 26 years…

Keynote
Educational Talk
Intermediate
Dragan Tomic has been a participant of the Data and AI industry his entire career (26 years). During his career, he participated in various technological waves (oltp, big data, cloud services and ai). He participated in building the SQL Server which is one of the leading database management systems. Also, he was at the forefront of the cloud services revolution where data oltp databases, data warehousing were reimagined in the cloud. He is currently leading a team at Databricks which is leading the Data and AI wave. In this presentation, Dragan will outline why database systems are the most beautiful part of computer science and offer some lessons about how to think about the current developments in the data and AI domain.
12:45

 –

13:30

Keynote: Key lessons from building Data and AI systems over the last 26 years…

Dragan Tomic
VP of Engineering and Site Lead, Belgrade @ Databricks

Lunch Break

13:30

 –

14:30

Lunch Break

AI on the Assembly Line: Real-Time Quality Control with Computer Vision

Computer Vision & Robotics
Technical talk, Business talk
Intermediate
Discover how Computer Vision and Machine Learning transform Quality Control processes in Manufacturing. In the session about AI on the Assembly Line, we will walk through the case of an AI-driven system capable of inspecting parts as they are produced, highlighting the business case, the real-time aspect, and the solution scalability.
14:30

 –

15:00

AI on the Assembly Line: Real-Time Quality Control with Computer Vision

Donovan Spronk & Vladislav Belov
Partner & AWS Alliance Leader @ Deloitte & Technical Lead @ Deloitte

Yandex Alice: music control with gestures

Computer Vision & Robotics
Technical talk
Intermediate to Advanced
When your voice assistant can also see, the range of ways you can interact with it expands significantly. One of its features is music control with gestures, not only with voice commands.
In my talk, I would like to tell you about the development of models for such a feature. We will focus on open-source data collecting ways, making lightweight models and optimizing them for a specific hardware.
15:00

 –

15:30

Yandex Alice: music control with gestures

Darya Vinogradova
Senior Computer Vision Engineer @ Yandex

Coffee Break

15:30

 –

16:00

Coffee Break

Structuring Unstructured Data to Boost Computer Vision and Generative AI Applications at Scale

Computer Vision & Robotics
Technical talk, Business talk
Intermediate
In the rapidly evolving fields of Computer Vision (CV) and Generative AI (GenAI), effectively structuring and managing unstructured data is a key challenge. This presentation will introduce innovative techniques aimed at tackling these challenges by enabling scalable data management and processing. We will explore how modern tools not only enhance data accessibility and integrity, but also seamlessly integrate with AI workflows to improve model training and inference capabilities. Attendees will gain insights into leveraging cutting-edge tools and methodologies to transform raw data into structured, actionable datasets that drive innovation in CV and GenAI projects.
16:00

 –

16:30

Structuring Unstructured Data to Boost Computer Vision and Generative AI Applications at Scale

Mikhail Rozhkov
Technical Product Manager @ Nebius

Manufacturing in 21st century: AI supported manual assembly

Computer Vision & Robotics
Technical talk
Beginner to Intermediate
Contrary to popular belief the majority of manufacturing is still dominated by humans. The idea that robotics and automation drive manufacturing is not the case, as evident on many shopfloors across the world. This is why we made our mission to apply AI to support the human skill on production lines, to ensure those skills are retained and reusable for the future.
The manufacturing industry is still prone to human errors which results in rework, potential losses for the company, increases waste and requires additional resources in time, materials and manpower. On top of that the work data in most cases not efficiently recorded, causing lack of traceability and makes it time consuming and difficult to backtrack anything that has been done or assembled.
All of the errors over the whole industry amount to almost 1 out of every 2 pounds of loss due to human error. The need to address this and help alleviate the issues is pressing. Precisely this need is what prompted Vision Intelligence to develop VIOLET AI system and assist manual assembly operators while also pushing manufacturing industry forward to align with industry 4.0. We developed the hardware and full stack systems which work in unison with human beings to ensure the products are deliver with the highest quality and safety in mind.
The most important innovation in the use of this technology is that rather than replacing humans with technology, this approach will essentially allow for the democratization of quality management and training in a manufacturing environment. The system is designed to involve and empower the worker.
16:30

 –

17:00

Manufacturing in 21st century: AI supported manual assembly

Krste Pangovski
CEO @ Vision Intelligence

Day Recap

17:00

 –

17:15

Day Recap

Ivo Andric A

Augmented Automation,

Big Data Analytics

Registration

09:00

 –

09:30

Registration

Data Analysis in the Age of Data Democratization

Augmented Analytics
Business talk
Intermediate
Everyone can access data. But can everyone access data analysis? This is the story of how we bridged the gap between non-technical users and data driven solutions.
09:30

 –

10:00

Data Analysis in the Age of Data Democratization

Ioan Enache
Data Specialist @ Data Revolt Agency & Programme Leader @ Digital Hub Academy

The (mis)use of GenAI in Data Analytics

Generative AI
Technical talk, Business talk
Beginner to Intermediate
The use of LLMs is currently abused in corporate environments, where stakeholders dont have neither the proper guidance to be able to avoid the pitfalls of hallucinations and other issues, nor the evaluation methodologies needed to avoid badly executed PoCs. In this talk we’ll go through what are the most common pitfalls and misuses of GenAI and how to avoid them, with an emphasis on Data Analytics for Marketing.
10:00

 –

10:30

The (mis)use of GenAI in Data Analytics

Guilherme Diaz-Berrio
Co-Founder @ Pinemarsh

From Insights to Impact: Transforming Automotive Infotainment with Data

Augmented Automation
Technical talk, Business talk
Intermediate
In the rapidly evolving automotive landscape, infotainment systems have become central to the user experience. To deliver the best user experience possible and drive innovation, Rivian has developed a robust analytics framework that enhances our software-defined vehicle paradigm.
This framework leverages data to:
● Enhance User Experience: By analyzing user interactions, we identify trends, preferences, and pain points, enabling continuous optimization of infotainment features.
● Drive Data-Informed Decision Making: The framework provides actionable insights for designers, product teams, business stakeholders, and engineers, allowing data-informed decisions that improve both user satisfaction and product performance.
● Strengthen Infrastructure: With advanced instrumentation, a seamless Vehicle-to-Cloud pipeline, and a scalable cloud infrastructure, we ensure reliable data collection, processing, and analysis.
This presentation will explore the key components of Rivian’s infotainment analytics framework, share real-world use cases, and highlight the value it brings to our organization.
10:30

 –

11:00

From Insights to Impact: Transforming Automotive Infotainment with Data

Nemanja Tiosavljevic
Senior Manager @ Rivian

Event Analytics – from unstructured to insightful

Augmented Automation
Technical Talk
Intermediate to Advanced
Has your CEO ever looked at a report you showed them, and said something looked off? Has your financial report ever come out with unexplainable expense origins? How often have you found yourself in an uncomfortable situation of struggling to find evidence to support the insights you presented to your customers? Can you relate to any of these scenarios? You’re not alone. In the modern business intelligence landscape, data is often sourced from multiple origins, necessitating the integration of this data into unified representations and entities that accurately reflect our system. This requirement introduces significant challenges, particularly in modeling the events generated by these entities. The provision of data points from diverse and expanding sources emphasizes the need to structure this data into cohesive frameworks that provide a comprehensive system view. This presentation showcases our technical approaches and insights on event modeling within the realm of business intelligence, calling attention to the detective work involved in reconstructing data narratives to derive enhanced insights. We will address complex issues such as data integration, event correlation, and entity resolution, presenting methodologies and best practices for optimizing these processes. Our discussion seeks to provide technical perspectives that enable organizations to manage and exploit their multi-source data adeptly. Addressing these challenges, we aim to advance the effectiveness of decision-making processes, improve operational efficiencies, and deepen our understanding of complex business environments.
11:00

 –

11:30

Event Analytics – from unstructured to insightful

Boris Paunovic & Igor Djordjevic
Tech Lead & Senior Data Analyst @ HTEC & Tech Lead & Senior Data Engineer @ HTEC

Coffee Break

11:30

 –

12:00

Coffee Break

Black

12:00

 –

12:45

Black

Black

12:45

 –

13:30

Black

Lunch Break

13:30

 –

14:30

Lunch Break

Data Platform Infrastructure at Rivian: A Deep Dive

Big Data Analytics
Technical talk
Advanced
Building and maintaining a reliable and efficient data platform requires a focus on performance and scalability while keeping costs under control. To achieve this goal we use different AWS and cloud-native tools and concepts. This talk provides a omprehensive overview of the platform infrastructure we run for serving some of the most demanding distributed systems used across the company. Moreover, it highlights the scaling challenges encountered with Druid and Flink and how we took advantage of Kubernetes to overcome them while keeping expenses low.
14:30

 –

15:00

Data Platform Infrastructure at Rivian: A Deep Dive

Veljko Tanjga
Tech Lead @ Rivian

AI-driven query plan optimization in big data processing tools

Big Data Analytics
Technical talk
Intermediate
In the world of big data where people frequently use tools like Hadoop, Hive, Spark and work with SQL-like queries in distributed environments, it is very important to estimate query execution costs correctly. Selecting a different query execution plan may have dramatic effects on performance.

The existing open source tools use query statistics + heuristics to build and compare query execution plans. The heuristics are not always accurate, and this is especially difficult for newly created data where query statistics is not yet available.

The commercial / cloud platform vendors (e.g. Databricks, Vertica) have additional sources of data to optimize their products compared to open source versions. It appears to be possible to train ML models on the accumulated data of many clients to better estimate the parameters on newly created distributed tables.

In the current talk, we shall discuss how much performance the AI-based cost estimation gives using TPC-DS and TPC-H benchmarks for measurement. The other interesting topic for discussion is where the open source tools could potentially gather the data to train comparable models.
15:00

 –

15:30

AI-driven query plan optimization in big data processing tools

Vadim Opolski
Senior Data Engineer & Global Data Chapter Lead @ Luxoft

Coffee Break

15:30

 –

16:00

Coffee Break

Harnessing Real-Time Power: Inside Sofascore’s Data-Driven Infrastructure

Big Data Analytics
Technical talk, Business talk
Intermediate to Advanced
Join us for a deep dive into Sofascore’s cutting-edge analytics pipeline, powering 25 million monthly users and processing over 1.5 PB of data. Discover how we leverage user events, Firebase, BigQuery, ClickHouse, NATS, and Kubernetes clusters to deliver real-time insights. From on-premise servers to Amazon S3 backups, see how AI-driven analytics unlocks the potential of big data, scaling infrastructure seamlessly for the ultimate sports experience.
16:00

 –

16:30

Harnessing Real-Time Power: Inside Sofascore’s Data-Driven Infrastructure

Karlo Knezevic
Head of AI @ Sofascore

Turn Data into Action: How data visualization aids decision intelligence

Big Data Analytics
Educational talk
Beginner to Intermediate
We live in an era where data is abundant, a large variety of models easily accessible and free, and yet, most machine learning projects are still set to fail. In this talk, we will delve into understanding, interpreting, and presenting data and explain the critical role of this process in making decisions prior to and throughout a project. We will have a look into the principles of data exploration aided by visual story-telling, a task which can strongly impact the decision making process and alter the course of a project. By looking at concrete examples, we will uncover how well-understood data and intuitive visualizations can reduce cognitive load of the audience and empower team leaders and stakeholders to make better decisions.
16:30

 –

17:00

Turn Data into Action: How data visualization aids decision intelligence

Sumeyye Suri
Data Analyst @ Sclable

Day Recap

17:00

 –

17:15

Day Recap

Lavender

AI & MLOps,

Deploying ML Models & Agents

Registration

09:00

 –

09:30

Registration

Effective Context Tuning Strategies for Retrieval-Augmented Generation (RAG) Systems

AI & MLOps
Technical talk, Research talk
Intermediate to Advanced
Retrieval-Augmented Generation (RAG) systems represent a cutting-edge approach to natural language processing, blending the strengths of retrieval-based models with generative architectures. Central to the effectiveness of RAG systems is the nuanced management of context. This abstract investigates various strategies for context tuning in RAG systems, ranging from the selection and weighting of relevant passages to the dynamic adjustment of context size. Through case studies and empirical analysis, we uncover insights into how different tuning strategies impact the quality and coherence of generated text. Join us as we navigate the landscape of context tuning in RAG systems, unlocking pathways to more effective and adaptable language generation.
09:30

 –

10:00

Effective Context Tuning Strategies for Retrieval-Augmented Generation (RAG) Systems

Sofia Konchakova
Researcher @ Humboldt University of Berlin

Data Science in the Fast Lane: Accelerating Model Deployment with Continuous Integration and MLOps

AI & MLOps
Technical talk
Intermediate
The data science landscape is constantly changing and the need for efficient, streamlined workflows is critical to getting machine learning models into production. Traditional methods of model development and deployment are often slow, manual, and prone to human error. This is where Continuous Integration (CI) and MLOps come into play – providing a framework for automating and accelerating model deployment. CI is a best practice borrowed from the software development world that automates all the steps required to build, test, review, and deploy code to your production environment. CI helps detect and resolve conflicts early and ensures the codebase remains stable. These same concepts can also be applied to data science workflows by automating tasks, such as, data ingestion, preprocessing, feature engineering, training, model validation, code review, and deployment. By seamlessly integrating CI into data science pipelines, teams can automate these essential tasks, allowing them to iterate through the model development process faster and deploy models into production with confidence. MLOps is an emerging but quickly maturing field that seeks to streamline the entire machine learning lifecycle – from model creation, testing, deployment, monitoring, and iteration. MLOps allows data scientists to deploy models into production faster, by providing the tooling needed to carry out efficient experimentation, scalability, and reproducibility. Combining the power of CI and MLOps unleashes the power of automated pipelines for data science and machine learning workflows. From this talk, attendees will: 1) Understand the principles and benefits of CI for data science; 2) Explore how the emerging field of MLOps complements CI to empower data science teams to be more efficient in deploying machine learning models; and 3) Discover practical strategies to implement CI and MLOps in data science workflows in any environment, via real-world examples.
10:00

 –

10:30

Data Science in the Fast Lane: Accelerating Model Deployment with Continuous Integration and MLOps

Kevin Dietz
Staff Data Scientist @ GitLab

DSPy: The End of Prompt Engineering as We Know It

AI & MLOps
Technical talk
Advanced
Traditional prompt engineering has become a critical bottleneck in LLM application development. Developers spend countless hours crafting, tweaking, and maintaining prompts through trial and error. This manual approach is not only time-consuming but also brittle – prompts that work today may fail tomorrow, and solutions that work with one model often break with another. As organizations scale their LLM applications, the complexity of managing hundreds of hand-tuned prompts becomes increasingly unsustainable, leading to inconsistent performance, high maintenance costs, and significant technical debt.

Enter DSPy – a revolutionary framework that fundamentally reimagines how we work with language models. By introducing automated prompt optimization and a modular programming approach, DSPy eliminates the need for manual prompt crafting. Instead of writing prompts, developers define their desired outcomes and metrics, allowing DSPy to automatically discover and optimize the most effective prompts and weights. The framework’s declarative programming model separates the business logic from the underlying prompt implementation, while its self-improving pipelines continuously tune performance based on real-world results.

In this talk, I will introduce DSPy’s core principles and demonstrate its use through practical examples, providing a good starting point to transform your LLM development workflow from manual prompt engineering to automated optimization.
10:30

 –

11:00

DSPy: The End of Prompt Engineering as We Know It

Boris Cergol
Head of AI @ Comtrade System Integration

Rivian Autonomy and AI: Building a Scalable Platform for Data and ML Workloads

AI & MLOps
Technical talk
Advanced
In this session we will explore how Rivian has developed a scalable platform to manage the complex data pipelines and ML workloads powering its advanced driver-assistance system (ADAS) platform. We’ll present the overall architecture and infrastructure of Rivian’s internal platform designed to streamline data processing and pipeline management. We will dive into the use of KubeRay, Kueue, and EKS for orchestrating ML workloads at scale, along with the custom infrastructure and tooling we built to support it.
11:00

 –

11:30

Rivian Autonomy and AI: Building a Scalable Platform for Data and ML Workloads

Aleksandar Cvejic
Team Lead, Data Platform & ML Infra, Autonomy & AI @ Rivian

Coffee Break

11:30

 –

12:00

Coffee Break

Black

12:00

 –

12:45

Black

Black

12:45

 –

13:30

Black

Lunch Break

13:30

 –

14:30

Lunch Break

Accelerating AI Innovation through Hyperscaler Platforms: Data Engineering and Data Science Perspective

Deploying ML Models & Agents
Technical talk
Intermediate to Advanced
This presentation deck explores how hyperscaler platforms enable faster development and scaling of AI models, particularly within the realm of data science. It focuses on the advantages of hyperscaler infrastructure, such as high performance, resource flexibility, and the ability to efficiently access, process, analize, train and deploy AI models on large datasets. The presentation also highlights best practices for leveraging cloud technologies and automatic scaling to accelerate innovation, reduce costs, and improve operational efficiency. Based on a practical example from a huge data source for mobile data customers behavior, model output provides insights for Telekom Serbia, aiming to enhance AI model development using hyperscaler platforms.
14:30

 –

15:00

Accelerating AI Innovation through Hyperscaler Platforms: Data Engineering and Data Science Perspective

Sinisa Arsic
Director of Department for Data, analytics and intelligent automation of business processes @ Telekom Srbija

Beyond Autocomplete: Local AI Code Completion Demystified

Deploying ML Models & Agents
Technical talk
Beginner to Intermediate
The talk is about the details and motivation behind Local AI Code Completion by JetBrains that was released on the 4th of April 2024. I’ll start by talking about the current state of the most popular AI Code Completion tools. We’ll discuss the pros and cons of running language models for code in the cloud versus on the user’s machine. Then we’ll go through the main components of AI Code Completion which is much more than just a language model. We will discuss our approach to offline and online evaluation of AI Code Completion, prompt construction, and smart suggestion filtering. Also, we’ll discuss the rapid progress in efficient language model inference that we’re witnessing nowadays and the perspectives that local AI has given this progress. Finally, I’ll share some insights about the size of the team that made Local AI Code Completion for JetBrains IDEs happen, the model sizes that are currently used in production, and the cost of the models’ training.
15:00

 –

15:30

Beyond Autocomplete: Local AI Code Completion Demystified

Daniel Savenkov
Senior ML Engineer @ JetBrains

Coffee Break

15:30

 –

16:00

Coffee Break

How to fail AI implementations

Deploying ML Models & Agents
Educational talk
Beginner to Intermediate
Many AI projects fail to achieve their goals. This talk dives into common pitfalls, exploring how to misstep in:
Scope Definition: Setting unrealistic expectations by promising the impossible, failing to identify the right problem for AI (e.g., applying AI to a task better suited for simpler rules), or neglecting to define clear success metrics that can be objectively measured.
Data Preparation: Using insufficient data that limits the AI model’s ability to learn, using irrelevant data that doesn’t pertain to the problem at hand, or using biased data that skews the model’s outputs and leads to unfair or discriminatory results.
Client Communication: Misunderstanding client needs by failing to gather in-depth requirements at the outset, failing to manage expectations by underestimating the complexity or timeline of the project, or neglecting to communicate challenges and roadblocks transparently throughout the development process.
Refine project scope, Improve data collection and cleaning, Foster clear communication with stakeholders, The talk will explore real-world examples and provide actionable takeaways to help you navigate the development process, turning potential failures into stepping stones for successful AI projects.
16:00

 –

16:30

How to fail AI implementations

Matevz Cerne
Head of fraud prevention and predictive analytics @ Sapphir

AI Agents: The Future of Autonomous Decision-Making in Business and Beyond

Deploying ML Models & Agents
Technical talk, Business talk
Intermediate
As we continue to push the boundaries of artificial intelligence, the concept of AI agents has emerged as a transformative force, offering a revolutionary approach to decision-making and automation. This presentation will explore the foundational principles of AI agents, their working mechanisms, and the evolution of this concept, leading to a new era of chained decision-making. Unlike traditional models that perform single tasks, AI agents unlock the potential for continuous, hierarchical decision processes, making them a game-changer in various industries.

Key Topics:
The Genesis of AI Agents:
Understanding the fundamental principles that guide AI agents: autonomy, perception, reasoning, and action.
The shift towards agent-based systems as a response to the need for more dynamic and adaptable AI solutions.

How AI Agents Work:
Breaking down the architecture: perception modules, reasoning engines, and action mechanisms.
The concept of goal-oriented behavior and how agents learn to optimize outcomes over time.

Real-world analogy: Comparing AI agents to human decision-making processes in complex environments.

Chained Decision-Making:
Unleashing the power of AI through chained decisions: how agents process sequences of tasks autonomously.
The advantages of continuous decision-making compared to single-execution models.
Examples from industries where this approach has transformed operations and outcomes.

Hierarchical AI Agents:
Introducing the concept of hierarchical agents: managing complexity through multi-layered decision-making structures.
The benefits of a hierarchical approach in dealing with large-scale problems and complex environments.

Potential use cases: From supply chain management to automated customer service, and beyond.

Some of the examples in Industry Applications:
Finance and Banking: Autonomous trading agents, fraud detection, and risk management.
Healthcare: Personalized patient care, automated diagnostics, and medical research.
Manufacturing: Smart factories, predictive maintenance, and logistics optimization.
Retail and E-commerce: Dynamic pricing, personalized recommendations, and supply chain automation.
Energy: Smart grids, resource management, and predictive analytics for sustainability.

The Future of AI Agents:
Envisioning the next generation of AI agents: Increased autonomy, self-learning capabilities, and collaborative agents.
Exploring the ethical and societal implications of deploying AI agents at scale.

Potential breakthroughs: AI agents as partners in creativity, innovation, and problem-solving.

This presentation aims to inspire and educate attendees on the groundbreaking potential of AI agents. By delving into their principles, mechanisms, and real-world applications, the session will offer a forward-looking perspective on how AI agents can redefine decision-making and automation across industries.

AI agents represent a paradigm shift in how we approach problem-solving and automation. By harnessing the power of chained decision-making and hierarchical structures, these agents have the potential to unlock unprecedented efficiencies and innovations across industries. This presentation will not only provide a deep dive into the workings of AI agents but also spark the imagination about what the future holds for this transformative technology.
16:30

 –

17:00

AI Agents: The Future of Autonomous Decision-Making in Business and Beyond

Lana Malic
Machine Learning Engineer @ Endava

Day Recap

17:00

 –

17:15

Day Recap

Tesla B – DSC:X

Data-Driven Culture in Organisations,

Applied AI in Industry

Registration

09:00

 –

09:30

Registration

Building a Data-Driven Culture: What the C-Suite Needs to Understand First

Data-Driven Culture in Organisations
Transformational talk
Intermediate
So, you prefer data, metrics, and facts over opinions, speculations, and anecdotes? Fair enough.

And you want to imprint that in your company’s DNA? Install it in your corporate OS? Fair enough.

As you launch, develop, and cultivate your data-driven culture, remember these important points:

– Hologram theory, individual gravity, and how corporate spacetime is bending.

– Barun Münchhausen pulled himself out of a mire by his own hair. So many companies are doing the same with their culture.

– Organization design has five components. Culture is only one of them; I would put it on the 5th place. If “”Culture eats strategy for breakfast”” is indeed correct it is only because the most important component is letting it do that.

If you understand these three talking points of our lecture, you are good to go. All the best in the building of your data-driven culture. You don’t need luck; you have knowledge and skills. But if you don’t understand it, you better come and hear it for yourself. I am guaranteeing you that these 30 minutes will be worth your time.
09:30

 –

10:00

Building a Data-Driven Culture: What the C-Suite Needs to Understand First

Fran Mikulicic
CEO @ FM Consulting

From Insight to Action — The “Secret Sauce” of Canada’s Top Public Sector Data Science Team

Data-Driven Culture in Organisations
Use case, Transformational talk
Intermediate
Discover the impactful strategy behind Edmonton’s Data Science and Research team, Canada’s leading public sector data science powerhouse. In ‘From Insight to Action,’ we reveal the ‘secret sauce’—a pragmatic approach formed during the COVID-19 crisis that parts ways with conventional AI strategies. Learn how we transform data into actionable solutions with a sharp focus on real-world problems, dedicated stakeholders, and an unwavering commitment to impactful change. This is your chance to gain insights from the team that’s redefining municipal data science and making a tangible difference.
10:00

 –

10:30

From Insight to Action — The “Secret Sauce” of Canada’s Top Public Sector Data Science Team

Kris Andreychuk
Manager, Data Science and Research @ City of Edmonton, Canada

BI Battle: Centralized Control vs. Decentralized Agility

Data-Driven Culture in Organisations
In a market that demands both continuous innovation and uncompromising quality, organizations face a critical choice in their business intelligence (BI) strategies: should they adopt a centralized approach to ensure consistency and control, or a decentralized, agile model that empowers departments to innovate and adapt? This session explores the strengths and weaknesses of both BI organizational frameworks, examining how each impacts operational cost, decision-making speed, and adaptability. Join us as we dive into the BI Battle and discover how to achieve the right balance between centralized control and decentralized agility for your organization.
10:30

 –

11:00

BI Battle: Centralized Control vs. Decentralized Agility

Stefan Stosic
Data Analytics Senior Manager @ NCR Atleos

Keynote: Journey to AI driven company

Keynote
Azfar will explore the journey to becoming an AI-driven company, emphasizing key strategies and challenges organizations face in leveraging AI for growth. He’ll highlight the importance of fostering a data-driven culture and collaboration between data science and business teams
11:00

 –

11:30

Keynote: Journey to AI driven company

Azfar Shah
Chief Strategy And Data Officer @ Yettel.Serbia

Coffee Break

11:30

 –

12:00

Coffee Break

Black

12:00

 –

12:45

Black

Keynote Fireside chat: Role of Leadership in Scaling AI across the Enterprise

Keynote
12:45

 –

13:30

Keynote Fireside chat: Role of Leadership in Scaling AI across the Enterprise

Natali Delic
Chief Strategy and Digital Officer and Executive Board Member @ Telekom Srbija

Lunch Break

13:30

 –

14:30

Lunch Break

Buidling cloud data platforms in enterprises

Applied AI in Industry
Cloud data platforms don’t come from thin air – they are created in long term strategic plan long before than people in operations and, usually, lower management, are aware what is going on. Lecture will give a short glimpse of what happens in world of corporate meeting rooms and what must happen that we start a cloud transformation project.
14:30

 –

15:00

Buidling cloud data platforms in enterprises

Josip Saban
Data consultant @ Sole proprietorship

Yottaanswers.com – bringing tens and hundreds of billions of best smart answers to users worldwide with thousand times fewer resources

Applied AI in Industry
The contemporary generative AI information revolution has been predicated upon absurd resource requirements of hundreds of billions dollars that only a few mega-tech companies and countries can contemplate spending. Even for them such insane levels of squander have resulted in losses and cash burn amounts that have never been seen in history. Yottaanswers.com is showing how such ridiculous requirements are but a mirage and an illusion and how to bring tens and hundreds of billions of best smart answers to users worldwide with thousand times fewer resources. In addition to making smart AI approachable to the rest of the world, our systems and technology will also show even to the biggest of contemporary AI players how to make a viable business out of modern AI by bringing profitability back into the equation.
15:00

 –

15:30

Yottaanswers.com – bringing tens and hundreds of billions of best smart answers to users worldwide with thousand times fewer resources

Borislav Agapiev
Founder @ YottaAnswers

Coffee Break

15:30

 –

16:00

Coffee Break

Challenges with AI implementation in organizations

Applied AI in Industry
16:00

 –

16:30

Challenges with AI implementation in organizations

Bojan Jovic
Director, Center of Excellence @ Avaya

Evaluating the Worth of Implementation: Insights from Process Mining Projects

Applied AI in Industry
Learn how to evaluate the potential impact and value of implementation projects within your organization.
Discover methods to predict benefits and measure success through real-world examples from process mining.
Explore best and worst practices for measuring and communicating value effectively.
16:30

 –

17:00

Evaluating the Worth of Implementation: Insights from Process Mining Projects

Natalia Vasileva
Senior PI Consultant @ T1A

Day Recap

17:00

 –

17:15

Day Recap

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