Sava Centar, Belgrade

20-24th November 2023

DSC SCHEDULE

DATA & AI TALKS

CONGRESS HALL B

Redefining User Experience with Large Language Models

Keynote
Keynote
09:15

 —

10:00

Redefining User Experience with Large Language Models

Igor Ilic
Principal Applied Scientist, Office Team @ Microsoft

Seeing Through the Lens of Transformers: A New Era in Computer Vision

Computer Vision & Robotics
Technical talk
Intermediate to Advanced
The digital world has witnessed a revolutionary transition as transformers, originally built for language tasks, pivot to the domain of vision. This talk offers a panoramic view of the transformative role of vision transformers, highlighting their unique architecture and capabilities. We’ll dissect the core components, explore the nuances of self-attention, and unveil the reasons behind their impressive performance in vision tasks.
Computer Vision & Robotics
10:00

 —

10:30

Seeing Through the Lens of Transformers: A New Era in Computer Vision

Ivan Biliskov
CEO @ Codeasy

Coffee break

10:30

 —

11:00

Coffee break

Automatic detection & geolocation of persons on UAV images to assist in search & rescue operations

Computer Vision & Robotics
Technical talk, Research talk
Intermediate
People enjoy spending time in the wilderness for a number of reasons. However, they occasionally get lost or injured, and their survival depends on being effectively found and rescued in the shortest possible time. After reporting the accident, a search and rescue operation begins, and all possible resources are activated. The inclusion of drones in SAR operations has made it possible to use computer vision methods to automatically detect people on aerial images. Deep learning models developed for the detection of persons in search and rescue operations will be presented, along with the challenges that exist in real-world scenarios.
Computer Vision & Robotics
11:00

 —

11:30

Automatic detection & geolocation of persons on UAV images to assist in search & rescue operations

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

Data & AI Talk

Computer Vision & Robotics
Computer Vision & Robotics
11:30

 —

12:00

Data & AI Talk

Finetuning Stable Diffusion with low-rank adaptation techniques

Computer Vision & Robotics
Technical talk
Intermediate
In this talk, I would like to tell the audience about the existing fine-tuning techniques for Stable Diffusion models including hypernetworks and low-rank adaptation techniques – their advantages and disadvantages and how they could be applied to all sorts of models not limited to image generation. I am sure that every listener would benefit from incorporating this approach into their downstream tasks.
Computer Vision & Robotics
11:30

 —

12:00

Finetuning Stable Diffusion with low-rank adaptation techniques

Alexander Kovalchuk
Deep Learning Engineer @ Prequel Inc.

Coffee break

12:00

 —

12:15

Coffee break

Beyond Deep Learning: The Cognitive AI Frontier

Keynote
In this talk, Rana will delve into the fascinating world of cognitive AI, shedding light on how this technology is pushing the boundaries of traditional AI models. Discover how Cognitive AI is revolutionizing industries, enhancing problem-solving capabilities, and offering a glimpse into the future of intelligent systems. Get ready to embark on a journey beyond deep learning and into the uncharted territory of AI’s transformative potential.
Keynote
12:15

 —

13:00

Beyond Deep Learning: The Cognitive AI Frontier

Rana Gujral
CEO @ Behavioral Signals

How to Build and Scale Ultra-Lean Gen AI Startups

Keynote
Join us for an electrifying keynote by Tarry, who brings a treasure trove of over 25 years of expertise to the stage. Dive into his groundbreaking theory that marries human creativity with turbocharged productivity – a blend that, while rare, sparkles when it occurs. In an era where high-performance computing and Generative AI are reshaping the norm, Tarry illustrates why this fusion is not just desirable but essential for team triumph. Known for his straight-talking, action-oriented approach, Tarry will also unveil the secrets behind his recent startup successes, where lean yet powerful teams have skyrocketed to over 1 million in revenues. Prepare to be inspired, enlightened, and propelled into a world where ingenuity meets efficiency!
Keynote
13:00

 —

13:45

How to Build and Scale Ultra-Lean Gen AI Startups

Tarry Singh
CEO @ DK AI Holding

Lunch break

13:45

 —

15:00

Lunch break

Robotic emotions

Computer Vision & Robotics
Technical talk, Research talk
Intermediate to Advanced
Psycho-emotional states play an important role in forming immediate reactions and coloring memories that play an important role in our further decisions in mammals. In this talk Max will present the approach to implement bio-inspired psycho-emotional states using dopamine, serotonin and noradrenaline sub-systems of a mouse brain. Max will demonstrate a bio-plausible model of a mouse brain and its application into robotic systems working in real time.
Computer Vision & Robotics
15:00

 —

15:30

Robotic emotions

Max Talanov
Senior Research Associate @ The Institute for Artificial Intelligence of Serbia

Engineering Intelligence through Agent Interaction

Computer Vision & Robotics
Technical talk
Intermediate to Advanced
Robotic process automation becomes agentic process automation in the realm of digital processes. Generative AI is expected to be the platform enabling replacement of menial manual labor, especially in digital workflows. However, engineering solutions that couple simpler components to manifest higher intelligence may require both cross-domain expertise and novel social and paradigmal considerations.
Computer Vision & Robotics
15:30

 —

16:00

Engineering Intelligence through Agent Interaction

Igor Sevo
Head of Artificial Intelligence @ HTEC

Coffee break

16:00

 —

16:30

Coffee break

How our neural networks find polygons to understand your documents better

Computer Vision & Robotics
Technical talk
Intermediate to Advanced
This presentation will discuss an important step in the document analysis pipeline: understanding the geometry of the document in the image. We will show pros and cons of several different deep learning approaches, ranging from simple ones like template matching to more advanced techniques such as differential rasterization and sampling-argmax. This talk may be useful for computer vision engineers working on tasks related to analyzing scenes with man-made objects, such as documents, aerial or satellite images, indoor scenes, and more.
Computer Vision & Robotics
16:30

 —

17:00

How our neural networks find polygons to understand your documents better

Boris Zimka
Principal Software Engineer @ ABBYY

Deciphering the AI Landscape: Business, Market Trends, and Emerging Risks

Closing Keynote
Technical talk, Business talk, Research talk
Intermediate to Advanced
I will unveil the findings of an exhaustive study on the current landscape of AI in business, its market dynamics, and the associated risks. This research, rooted in the Natural Language Processing (NLP) analysis of media discourse, utilised leading language models (LLMs) and drew from a variety of data sources. Beyond the empirical picture, I will dive into an analysis of the various risk levels inherent to the AI business. Additionally, I will provide insights into the potential direction of the AI market in the near to medium-term future. The primary objective is to clarify our present understanding and control of the AI business, while also highlighting the uncharted areas we might encounter in the future.
Closing Keynote
17:00

 —

17:45

Deciphering the AI Landscape: Business, Market Trends, and Emerging Risks

Goran S. Milovanovic
Lead Data Scientist @ Smartocto, Owner @ Data Kolektiv

Day Recap

18:00

 —

18:15

Day Recap

CONGRESS HALL A

Break

09:15

 —

09:30

Break

ML in Production on Databricks

AI & MLOps
Technical talk
Intermediate
Deploying ML models in production, with or without CI/CD, is significantly more complicated than deploying traditional applications. That is mainly because ML models do not just consist of the code used for their training, but they also depend on the data they are trained on and on the supporting code. Monitoring ML models also adds additional complexity beyond what is usually done for traditional applications. This talk will cover these problems and best practices for solving them, with special focus on how it’s done on the Databricks platform.
AI & MLOps
09:30

 —

10:00

ML in Production on Databricks

Petar Zecevic
Senior Principal Consultant @ Poslovna Inteligencija

Using AI for cheaper and more efficient solar power

AI & MLOps
Technical talk, Business talk, Research talk
Intermediate to Advanced
One of the global sustainability challenges nowadays is to have a planet powered by clean, affordable and reliable energy. The imperative would be to empower people to switch to affordable, reliable clean energy solutions with trusted resources. This spans from solar and batteries to heat pumps and electric vehicles, and many more. We have been building tools for image processing, clever matchings, and automated advising all with one goal to reach financial and environmental benefits of local solar power.
AI & MLOps
10:00

 —

10:30

Using AI for cheaper and more efficient solar power

Milan Bradonjic
Senior Software Engineer @ Schneider Electric

Coffee break

10:30

 —

11:00

Coffee break

DataOps & MLOps in Bosch: NLP Development in Kubernetes made easy

AI & MLOps
Technical talk
Intermediate to Advanced
DevOps practices are now extending into data processing and machine learning lifecycles, aiming to transition from proof-of-concept to scalable production systems. Our in-house solution leverages Kubernetes for seamless development and staging for developers and NLP researchers, with features like pipeline orchestration, task parallelization, and scheduling. This presentation showcases our comprehensive DataOps and MLOps system design, highlighting tools like Argo Workflows, MLflow, Seldon, and infrastructure integration with Terraform and Flux, all within the AWS cloud environment.
AI & MLOps
11:00

 —

11:30

DataOps & MLOps in Bosch: NLP Development in Kubernetes made easy

Igor Stranjanac & Filip Novovic
AI & Machine Learning Engineer @ Bosch & Senior Data Engineer @ Bosch

Mastering AI Issue Debugging: A Data Scientist’s Guide

AI & MLOps
Technical talk
Intermediate
Ever felt the frustration of tackling AI product glitches? You’re not alone. The process usually involves sifting through multiple data sources and logs, which can be quite a headache. But fear not, there’s a simpler way. Let’s explore a solution that offers a unified view of AI services-related info from log monitoring and data storage services. This talk dives into how to develop a user-friendly interface that benefits Data Scientists, support teams, testers, and developers in efficiently troubleshooting and debugging. By minimizing resolution time, we’re set to elevate the performance and quality of AI products.
AI & MLOps
11:30

 —

12:00

Mastering AI Issue Debugging: A Data Scientist’s Guide

Liliya Akhtyamova
Sr. Data Scientist / ML Engineer @ iCIMS

Coffee Break

12:00

 —

12:15

Coffee Break

Break Annex 1

12:15

 —

13:00

Break Annex 1

From Psychohistory to Curious Machines

Keynote Fireside Chat
Psychohistory of Today — Asimov’s Foundation and background metaphor for the rest of the presentation etc.

The Rise of Strategic Foresight

AI-powered Foresight w/ Nextatlas — how we work at Nextatlas, early adopter identification as our approach to data sources, how we built it, a brief showcase of our generative capabilities — Generate — in a short screencast + presenting the newest trends

Curious Machines
Keynote Fireside Chat
13:00

 —

13:45

From Psychohistory to Curious Machines

Luca Morena, Moderated by Jasmina Nikolic
Co-Founder & CEO @ Nextatlas, Client Service Director @ New Moment

Lunch break

13:45

 —

15:00

Lunch break

MetaHuman Framework: Use of simulations and ML body deformer in Unreal Engine

ML Application
Technical talk
Advanced
This talk will introduce a data powered product suite from Epic Games developed to support creation, animation and use of convincing synthetic humans branded as MetaHumans. We will then further focus on the proof of concept demo presented at Game Developers Conference in 2023, where traditional simulations derived body deformation data has been applied to train performant ML body deformer that runs in real-time on consumer grade hardware (in this case Play Station 5).
ML Application
15:00

 —

15:30

MetaHuman Framework: Use of simulations and ML body deformer in Unreal Engine

Uros Sikimic
Business Development Lead @ 3lateral

Cognitive Agents for Primary School Education: A Visual Approach

ML Application
Research talk
Intermediate
In this talk, work-in-progress results of a research project, which aims to enhance the learning outcomes of primary school students by using cognitive agents as interactive tutors, are presented. The project involves students designing their own cognitive agents by drawing and naming them, implemented by the project team. The cognitive agents can answer spoken questions from the curriculum, providing auditory feedback. How can this approach foster creativity, motivation, and self-regulation among learners, as well as improve their understanding of core artificial intelligence concepts. Further discussion on the challenges and opportunities of implementing this project in different educational contexts is provided.
ML Application
15:30

 —

16:00

Cognitive Agents for Primary School Education: A Visual Approach

Bogdan Okresa Djuric
Senior Teaching Assistant @ University of Zagreb, Faculty of Organization and Informatics, AI Lab

Coffee break

16:00

 —

16:30

Coffee break

Video transcription with deep learning module for recognizing sounds

ML Application
Technical talk, Business talk
Intermediate
How to get a good transcript and then supplement it with information about the background sounds? The leitmotif will be the use of Azure Video Indexer cognitive technology – a tool that enables transcription – and its integration with custom code written in Azure Databricks. During the lecture, I will show the integration of two technologies and present the approach to machine learning – I will review the available models for audio transcription, but also for image classification – which is especially useful for visual audio representation – spectrograms.
ML Application
16:30

 —

17:00

Video transcription with deep learning module for recognizing sounds

Pawel Ekk-Cierniakowski
Data Science Practice Lead @ SoftwareOne

How can Facebook data and machine learning algorithms predict political elections?

ML Application
Research talk
Intermediate
This talk explores the application of machine learning algorithms on the social network Facebook users’ activity data to develop predictive models of election outcomes. The main goal is to compare their accuracy and reliability with the models obtained by traditional public opinion polls. Four different approaches to machine learning are used to develop predictive models: an approach based on: error, information, similarity, and probability. The most effective approach is revealed. Obtaining equally effective predictive models with faster and easier access to data makes a significant scientific and social contribution to research in this area.
ML Application
17:00

 —

17:30

How can Facebook data and machine learning algorithms predict political elections?

Alen Kisic
CEO @ ZONA SJEVER d.o.o.

Break Annex1

17:30

 —

18:15

Break Annex1

CONGRESS HALL C

Break Annex 2

09:15

 —

09:30

Break Annex 2

Innovations in Proptech: Data-Driven Strategies for Distressed European Properties

Decision Intelligence
Business talk
Intermediate
Focusing on VendueTech’s aggregation of data from 250,000+ public auctions in the EU, this talk delves into innovative data-driven strategies for distressed properties, aiming to democratize real estate investment.
Decision Intelligence
09:30

 —

10:00

Innovations in Proptech: Data-Driven Strategies for Distressed European Properties

Ivan Livic
Co-founder & CEO @ Vendue Tech

Why new products still fail. Unlocking the Power of Sales and Marketing in post-human times.

Decision Intelligence
Business talk
Beginner to Intermediate
This engaging talk will explore the paradox of new product failure in the AI era. We will delve into the transformative impact of AI on sales and marketing, highlighting the shift towards customer-centric design and AI-driven strategies. The discussion will encompass the role of AI in e.g., predictive analytics or audience segmentation and the role of human communities for product success. We will also examine how AI enhances customer-centric design in sales funnels, leading to hyper-personalized content and offerings. The talk concludes with a look at future trends, emphasizing the continued importance of human-centric design while pushing AI integration in the post-human era.
Decision Intelligence
10:00

 —

10:30

Why new products still fail. Unlocking the Power of Sales and Marketing in post-human times.

Philipe Reinisch
Managing Director @ SILKROAD 4.0

Enhancement of Demand Forecasting and Supply Chain Planning product via GPT

Decision Intelligence
Technical talk, Business talk
Intermediate
Explore the transformative power of AI in supply chain optimization with the BE-terna Demand Forecasting and Supply Chain Planning platform. This innovative solution leverages cutting-edge AI algorithms for demand forecasting and inventory management, tailored to various item types and designed to streamline operations. Learn how this platform employs cloud-based deployments for cost-effectiveness, security, and seamless user experience. Moreover, in this talk we will discover how GPT is seamlessly integrated into the system. The chatbot that acts as a friendly guide, demystifying the complex algorithms behind stock recommendations and fostering transparency. It enables users of all technical backgrounds to interact with the AI, ask questions, and gain a deeper understanding of the logic driving optimization recommendations. Join us for an insightful discussion on making AI-driven recommendations accessible and comprehensible to a wider audience, ultimately improving trust and decision-making in supply chain management.
Decision Intelligence
11:00

 —

11:30

Enhancement of Demand Forecasting and Supply Chain Planning product via GPT

Nino Pozar
Data Scientist & Product Owner @ Be-terna

AI & CBS Made Complex Market and Consumer Research Analytics – Easy

Decision Intelligence
Technical talk, Business talk
Beginner
Our advanced analytics platform, Brand Pulse, is designed to be intuitive and user-friendly for marketing and sales data analysts and researchers. However, it can be complex for business users that are not so close to data and statistics. To solve this issue, we’ve integrated it with an AI-powered chatbot that can translate a business request into a Brand Pulse request, which connects statistical tools to provide usable insights. Now, users can communicate their requests in plain English and still get results of quite complex analysis in an easily readable format. Join us in this demo as we guide you through the integration of Azure Cognitive Search and Azure Open AI into Brand Pulse, which made the platform easy-to-use for everyone.
Decision Intelligence
11:30

 —

12:00

AI & CBS Made Complex Market and Consumer Research Analytics – Easy

Milan Bjeloglav & Vuk Vujacic
CEO @ CBS Systems & Analytics Consultant @ CBS Systems

Coffee break

12:00

 —

12:15

Coffee break

Break Annex 2

12:15

 —

13:00

Break Annex 2

LLM use in business transformation

Panel Discussion
Panel Discussion
13:00

 —

13:45

LLM use in business transformation

Jelena Djurovic & Miroslav Milivojevic & Stefan Dejanovic, Moderated by Sanda Martincic-Ipsic
Partnerships Manager Serbia & SEE @ Payoneer & Senior iOS Engineer @ LiquidText & Co-Founder & CTO @ GapApp, Full Professor @ Faculty of Informatics and Digital Technologies, University of Rijeka

Lunch break

13:45

 —

15:00

Lunch break

Scalable and Interoperable Data Flow Management for Digital Twin Frameworks in Smart Farming

Big Data Analytics
Technical talk, Research talk
Intermediate
In recent years, there has been a significant increase in the use of Smart Farming Technologies (SFTs), which are seen as key enablers in farm management for crop monitoring and reduction of chemical use. This presentation will cover a key component for the advancement of such systems – a data infrastructure which offers semantic and syntactic interoperability. Through the utilization of ontologies and smart data models in the agricultural domain, this kind of infrastructure can support actionable digital twins and advance farming capabilities.
Big Data Analytics
15:00

 —

15:30

Scalable and Interoperable Data Flow Management for Digital Twin Frameworks in Smart Farming

Mihailo Ilic
Teaching Assistant and Software Engineer @ VizLore Foundation

Implementing a Robust CI Workflow with dbt for Optimizing Data Warehouse Quality

Big Data Analytics
Technical talk, Educational talk
Intermediate
In this talk, I will discuss how to implement a robust CI (Continuous Integration) workflow with dbt (Data Build Tool) to optimize data warehouse quality. The implementation of a CI workflow can streamline collaboration and improve the quality of the data warehouse by catching errors early in the development process. By leveraging dbt’s modular approach and test-driven development practices, the CI workflow can help data teams ensure the accuracy and reliability of their data. Attendees will learn best practices for implementing a CI workflow with dbt and how to optimize their data warehousing quality
Big Data Analytics
15:30

 —

16:00

Implementing a Robust CI Workflow with dbt for Optimizing Data Warehouse Quality

Matteo Molteni
Data engineer @ Kognic

Coffee break

16:00

 —

16:30

Coffee break

Your data lakes need architecture too!

Big Data Analytics
Technical talk, Research talk
Beginner to Intermediate
Data Lakes became a dominant path in the Big Data analytics when it comes to storage and processing, especially recently by having them available in the cloud. More and more enterprises rely on them, but by throwing a huge volumes of data and without a proper architecture and data governance, they soon become obsolete data swamps. In order to avoid that, different approaches to setting a data lake architecture have been presented – and in this talk, we will go through each one of them and present their pros, cons and examples when they should be used.
Big Data Analytics
16:30

 —

17:00

Your data lakes need architecture too!

Tomislav Hlupic
Senior consultant @ Poslovna inteligencija

Real-time analytics at IoT scale

Big Data Analytics
Technical talk, Business talk, Research talk
Intermediate to Advanced
Rivian makes adventurous electric vehicles with a mission of a sustainable planet and keeping the world adventurous forever. Rivian’s vehicles are born in the cloud and embody tenets of a software defined vehicle, where not only the user accessible features such as infotainment are software driven and updated, but also internals aspects such as vehicle dynamics. Real-time instrumentation and telemetry are the key underpinnings that make all this possible. Rivian has built a cutting-edge Real-time stack using a combination of open-source technologies like Kafka, Flink and Druid and in house services. This talk will go into how these are combined and leveraged to deliver real-time analytics.
Big Data Analytics
17:00

 —

17:30

Real-time analytics at IoT scale

Pramod Immaneni
Principal Architect, Data Platform @ Rivian

MR 7

Break DSC:X

09:15

 —

09:30

Break DSC:X

How Data Hub flies above the clouds!

Data in Cloud
“Our central cloud data platform called Data Hub is part of our mission democratizing data analytics. It provides easy data access for data analysts and scientist with all required capabilities for data visualization, advanced analytics (ML, AI, ..), data governance and data cataloging. We are demonstrating our current setup based on the Databricks Lakehouse approach and taking you into a journey of how you can easily integrate, monitor and govern your data.”
Data in Cloud
09:15

 —

10:00

How Data Hub flies above the clouds!

Strahinja Miljkovic & Nikolaos Servos
Data Engineer @ Bosch & Technical Lead PT Data Hub @ Bosch

Data Lakehouse: Who, Why and How

Data in Cloud
We will dive into modern data management approaches that have become prevalent and popular across many industries, built on top of good old data lakes: Lakehouse. Here are some of the most common problems that are being solved with this novel approach: Data Silos Demolished: Discover how organizations are breaking down data silos that have plagued them for decades, unifying structured and unstructured data from diverse sources. Inefficient Data Processing: We’ll unveil real-world examples of how inefficient data processing can grind productivity to a halt and explore how Data Lakehouses provide a powerful solution while improving governance and security. Real-time Analytics: Learn how modern businesses are striving to achieve real-time analytics and the role Data Lakehouses play in achieving this. Have one data copy that will serve BI, Reporting, and ML workloads
Data in Cloud
10:00

 —

10:30

Data Lakehouse: Who, Why and How

Milos Solujic
CoFounder & CIO @ Opcom.io

Coffee break

10:30

 —

11:00

Coffee break

Cloud warehouse monitoring ( Snowflake case study )

Data in Cloud
Use case
Intermediate to Advance
Cloud data warehouses promise speed, scalability and data security/governance, but they rarely talk about monitoring and cost increase when we go into production and we start to process large datasets. This short lecture would show basics of user and cost monitoring, using Snowflake as leading cloud data provider, but demonstrated principles are applicable to any platform, and will show importance of implementing and using a monitoring system, both from financial and security aspect. You don’t need Snowflake experience to attend the class, just some interest in the topic of cloud governance.
Data in Cloud
11:00

 —

11:30

Cloud warehouse monitoring ( Snowflake case study )

Josip Saban
CEO, Meridian Data

AI Ops and its challenges

Data in Cloud
Transformational talk
Intermediate
We will start with the problems that have arisen in BigData with the increase in the volume of equipment. Let’s touch on what solutions are used for this (CMDB, Observability, configuration management tools, etc.). Then we will turn to the experience of solving the same difficulties long before the very concept of BigData appeared and look at a possible solution taking into account this experience
Data in Cloud
11:30

 —

12:00

AI Ops and its challenges

Alexander Buyanov
DevOps Evangelist @ foliage.dev | MTS

Data Lakehouse: Who, Why and How

Data in Cloud
We will dive into modern data management approaches that have become prevalent and popular across many industries, built on top of good old data lakes: Lakehouse. Here are some of the most common problems that are being solved with this novel approach: Data Silos Demolished: Discover how organizations are breaking down data silos that have plagued them for decades, unifying structured and unstructured data from diverse sources. Inefficient Data Processing: We’ll unveil real-world examples of how inefficient data processing can grind productivity to a halt and explore how Data Lakehouses provide a powerful solution while improving governance and security. Real-time Analytics: Learn how modern businesses are striving to achieve real-time analytics and the role Data Lakehouses play in achieving this. Have one data copy that will serve BI, Reporting, and ML workloads
Data in Cloud
11:30

 —

12:00

Data Lakehouse: Who, Why and How

Milos Solujic
CoFounder & CIO @ Opcom.io

Coffee break

12:00

 —

12:15

Coffee break

How to Fail with AI – and How to Succeed

Keynote
Transformational talk
Beginner to Intermediate
Getting value from AI is hard. In this presentation, you’ll discover the typical anti-patterns that lead to AI project failures and budget drains. From there, we’ll shift the focus to proven strategies, established best practices, and key insights to ensure your AI initiatives will actually get off the ground and produce meaningful value for your business.
Keynote
12:15

 —

13:00

How to Fail with AI – and How to Succeed

Tobias Zwingmann
Author & Managing Partner, RAPYD.AI

Building a Modern Data Platform for Software Defined Electric Vehicles

Keynote
Explore Rivian’s innovative data platform, a backbone for software-defined electric vehicles. This architecture efficiently collects telemetry data from the EV fleet and charger network, informing engineering, design, and reliability. Real-time processing identifies anomalies and serves data to an API layer that enables internal diagnostics and mobile clients. A data lake enables historical fleet-wide analytics and empowers data science and machine learning. This presentation unveils the platform’s key role in advancing software-defined EVs, emphasizing its architectural principles and analytical potential for industry transformation.
Keynote
13:00

 —

13:45

Building a Modern Data Platform for Software Defined Electric Vehicles

Jason Shiverick & Venkat Sivasubramanian
Principal Data Scientist @ Rivian & Senior Director IoT, Data, Analytics, ML @ Rivian

Lunch break

13:45

 —

15:00

Lunch break

Zero-party data giving AI the power of “why”

Applied AI in Industry
Use case, Transformational talk
Beginner to Intermediate
A talk on the traps faced by marketers in using data for activation and decision making currently and why Zero-party data, in particular, solves many of these problems. ZPD is fundamental in answering “why” people do what they do and this gives AI an edge in being able to not only understand users better, but to know how to deliver a treatment that is more likely to resonate.
Applied AI in Industry
15:00

 —

15:30

Zero-party data giving AI the power of “why”

Umar Akhtar
Co-founder | Chief Marketing Officer, Qudo

Eye-tracking and AI: Unveiling the Limitations and Synergies

Applied AI in Industry
Use case, Panel with the solution provider and the client
Intermediate to Advance
Applied AI in Industry
15:30

 —

16:00

Eye-tracking and AI: Unveiling the Limitations and Synergies

Milica Kovac
Researcher and New Business Development Manager, EyeSee

Coffee break

16:00

 —

16:30

Coffee break

Process mining from the inside

Applied AI in Industry
Transformational talk
Intermediate to Advance
Process mining approach can and should be a part of our everyday toolkit for managing our key business processes and making decisions – just as BI and Excel. Let me cover the basic ideas and principles of the technology, the cases and possible results – I’d like to tell what process mining is and how it could be used. And also we’ll talk about the approach and some tips that can make the adoption of process mining as smooth as possible.
Applied AI in Industry
16:30

 —

17:00

Process mining from the inside

Oksana Astapova
Head of process Intelligence @ T1A

Unlocking Geospatial Data Science: A Strategic Overview

Applied AI in Industry
Use case, Transformational talk
Beginner
In this talk, we will give a concise overview of the current geospatial data science and location intelligence landscape from C-level perspective to hands-on execution. We will cover i) trending data sources and types, including availability and pitfalls; ii) explore potential use cases for several industry segments; and iii) dive into a practical example from data collection to clear insights.
Applied AI in Industry
17:00

 —

17:30

Unlocking Geospatial Data Science: A Strategic Overview

Milan Janosov
Cheif Data Scientist @ Baoba insurance

BREAKOUT SESSIONS

MR 8

Q-Tangled (Quantum Computing)

09:30 – 10:00   Quantum Computing with Qiskit

Sasa Lazarevic
Data & AI Architect / IBM Quantum Ambassador @ IBM



10:00 – 10:30   Finding the winning quantum algorithm

Petar Korponaic
Founder and CTO @ Quantastica

In order to solve hard computational problems facing modern industry, having a fault-tolerant quantum computer alone is not enough. You will also need an algorithm, and that is not trivial.



10:30 – 11:00   Quantum and Today’s security with Quantum

Ales Gros
EET Partner Technical Specialist Manager and CEE Architect Community Leader @ IBM



11:00 – 11:30   Academia’s Role in Advancing Quantum Technologies

Aleksandra Gocanin
Assistant Professor @ University of Belgrade, Faculty of Physics

Academia plays a pivotal role in advancing quantum technologies, delving into the intricate realms of quantum mechanics and computing. Researchers explore the fundamental principles that underpin quantum phenomena, paving the way for groundbreaking applications. The synergy between academic institutions and quantum technology companies fosters innovation, pushing the boundaries of what’s possible in computing, communication, and sensing. As academia unravels the mysteries of quantum mechanics, it not only expands our understanding of the universe but also propels us toward a future where quantum technologies redefine the limits of computation and communication.



11:30 – 11:45   Coffee break



11:45 – 12:15   Big photons for big systems: Mid-infrared quantum optics in silicon

Marija Radulovic
PhD candidate, Advisor @ Quantum Engineering Centre for Doctoral Training, University of Bristo

We exploit the benefits of 2um wavelength in silicon-based integrated photonics, leveraging reduced nonlinear loss and increased linear gains. We focus on developing cointegrated large-scale systems for photon-based quantum computing.



12:15 – 13:00   Panel discussion

Sasa Lazarevic & Petar Korponaic & Ales Gros, Aleksandra Gocanin & Marija Radulovic
Data & AI Architect / IBM Quantum Ambassador @ IBM & Founder and CTO @ Quantastica & EET Partner Technical Specialist Manager and CEE Architect Community Leader @ IBM & Assistant Professor @ University of Belgrade, Faculty of Physics & PhD candidate, Advisor @ Quantum Engineering Centre for Doctoral Training, University of Bristo



09:30

 —

13:00

Q-Tangled (Quantum Computing)

Sasa Lazarevic, Petar Korponaic, Ales Gros, Aleksandra Gocanin, Marija Radulovic

Data governance in large Enterprises

15:00 – 16:30   Panel: Data governance in large Enterprises

Biljana Atanaskovic & Petra Hajdukovic & Miha Kovac & Andrej Hudoklin & Aleksandar Nedeljkovic
Head of Data Management and Analytics @ Nelt Group & BI & Analytics Director @ Fortenova Group & Executive Director Data Services @ Dormeo Home & Executive & Global Advisory and Implementation Services @ DataOrbis Slovenia & Chief Business Officer @ Mainstream
15:00

 —

16:30

Data governance in large Enterprises

Biljana Atanaskovic, Petra Hajdukovic, Miha Kovac, Andrej Hudoklin, Aleksandar Nedeljkovic

MR 10

End-to-end ML

9:30 – 9:45   Why typical DevOps pipelines are useless in ML?

Alexander Buyanov
DevOps Evangelist @ foliage.dev | MTS

“I’d like to involve the audience to conversation about problems with standard pipelines in ML: Why it can’t work well? Which decisions we have in industry for this really specific case? What about regulation in this area? And, finally, how all of this related with AIOps?”



9:45 – 10:00   Empowering Business with Pepsico’s Advanced ML Platform

Milos Grubjesic
ML Engineer @ PepsiCo

Discover the transformative power of machine learning in business with Milos. Dive into Pepsico ecommerce journey of building a robust ML platform. This talk sheds light on the decision-making process, the platform’s core features, and its scalable architecture on Azure and Kubeflow. Learn how such platforms can revolutionize data science workflows, from data access to model monitoring

10:00 – 10:15  Scaling AI: From Prototypes to Millions of Users

Nikola Milojevic
Co-Founder @NeuralFactory | ML Engineer @GridDynamics

Scaling AI: From Prototypes to Millions of Users is a deep dive into how we take a simple AI idea and grow it into a powerhouse used by millions. This talk will highlight the importance of planning for growth from the start—ensuring the AI we build can handle the jump from a handful to a multitude without stumbling. Learn the secrets to avoiding the pitfalls that keep most AI projects from making it big. Whether you’re an AI enthusiast or an industry professional, discover how to craft AI that’s ready for the spotlight from day one.



10:15 – 10:30  Leveraging LLMs for mapping of sport events

Stefano Gemin, Lidija Jovanovska
Data Science Team Lead @ Sportradar, Data Scientist @ Sportradar

Since Sportradar’s early days, when the main business was detecting arbitrage in sport betting, one of the main tasks has been to map sport events across different betting operators, federations, and sport news providers. Such mappings need to happen fast and with a very high degree of accuracy, with main challenges to that being the size of data, different languages and formats, but also the usage of abbreviations or nicknames, quite common in sports. Although technologies like Elasticsearch played a pivotal role in enhancing this process, the utilization of Large Language Models has demonstrated to be a significant improvement.



10:30 – 10:45  From Prompt to Product: Microsoft 365 Copilot

Tamara Stankovic
Senior Applied Science Manager @ Microsoft

In this talk, we present our journey of building Microsoft 365 Copilot in Word, a new set features designed to help users improve their authoring and consumption experience. We dig deeper into the Large Language Models (LLMs), prompt engineering, evaluation methods and Responsible AI mitigations, needed to build this feature and release it to millions of Microsoft users.



10:45 – 11:15  Panel discussion

Alexander Buyanov, Milos Grubjesic, Nikola Milojevic, Stefano Gemin, Lidija Jovanovska, Tamara Stankovic
DevOps Evangelist @ foliage.dev | MTS, ML Engineer @ PepsiCo, Co-Founder @NeuralFactory | ML Engineer @GridDynamics, Data Science Team Lead @ Sportradar, Data Scientist @ Sportradar, Senior Applied Science Manager @ Microsoft
09:30

 —

11:15

End-to-end ML

Alexander Buyanov & Milos Grubjesic & Nikola Milojevic & Stefano Gemin & Lidija Jovanovska & Tamara Stankovic

NLP for “Small-Country” Languages

11:30 – 11:45  Large Language Models for Serbian

Mihailo Skoric
Research assistant and software engineer @ University of Belgrade, Faculty of mining and geology

This talk will cover the range of published large language models (mainly RoBERTa and GPT-2), that were pre-trained specially for the Serbian language. Some of their characteristics will be highlighted, namely sizes and datasets used for the training of tokenizers and the pre-training of the models themselves. Lastly, the results of a crude comparison study on several tasks, such as text embedding and masked language modelling will be briefly presented.



11:45 – 12:00  Empowering Small Languages: SrBERTa – A Legal NLP Model for Serbian

Nemanja Petrovic
Tech lead @ Badin Soft

In this session, we delve into the world of NLP models designed for smaller languages, focusing on Serbian. I’ll introduce SrBERTa, a robust AI model finely tuned for Serbian legal texts. We’ll explore the unique challenges small languages face in NLP and how SrBERTa addresses them. Discover how this model improves legal document analysis and enhances accessibility to legal information in Serbian. Join us to see how SrBERTa is transforming the law and language tech field, emphasizing the untapped potential of NLP in underrepresented languages. Whether you’re a linguist, legal professional, or simply curious, there’s something here for everyone. Let’s harness the power of small languages together!



12:00 – 12:15  Machine Translation: should we trust it?

Zsolt Feleki
Assistant Professor @ University of Pecs

Should we trust Machine Translation (MT)?We all know the biblical story of the Tower of Babel, when God confused the people by creating many languages. Luckily though, God gave us the ability to learn other languages, so it is impossible for two humans not to communicate. In my presentation I show several examples of Machine Translation Pitfalls, when confusion is foredoomed. The talk will be highly interactive and we should find some answers to the questions:1. How does actually MT function?2. Is it somehow possible to improve MT in order to avoid misunderstanding?3. Are there areas where MT can be trusted? Or do we need MT at all?



12:15 – 12:30  Development of open NLP/NLU tools and resources for the Serbian language

Slobodan Markovic
Digital Advisor @ UNDP

I’ll be talking about an initiative I helped launch, aimed at developing open NLP/NLU tools and resources for the Serbian language. I’ll also emphasize why both the tech industry and governments need to significantly increase their investments to ensure that low-resource languages are well-represented in the ongoing AI revolution.



12:30 – 13:00  Panel discussion

Mihailo Skoric, Nemanja Petrovic, Zsolt Feleki, Slobodan Markovic
Research assistant and software engineer @ University of Belgrade, Faculty of mining and geology, Tech lead @ Badin Soft, Assistant Professor @ University of Pecs, Digital Advisor @ UNDP
11:30

 —

13:00

NLP for “Small-Country” Languages

Mihailo Skoric & Nemanja Petrovic & Zsolt Feleki & Slobodan Markovic

Intersection of Data & Blockchain

14:30 – 14:45   Future of Education with Blockchain

Djordje Petrovic
Professor @ George Brown college



14:45 – 15:00   Applied Decentralization / Decentralized or not really?

Marko Lazic
Blockchain Engineer @ DcentraLab

Talk about the importance and application of decentralization. Though decentralization is the most important feature of the blockchain it may not really be the focus of the most projects. How decentralized the most famous protocols truly are? How does participation in decentralized systems affect the persons life and how can we work towards establishing proper decentralization? Join the talk and find out 🙂



15:00 – 15:15   Making Blockchain Accessible Through Conversational AI

Johannes Pfeffer
Founder @ Hiero AI

Navigating blockchain transactions can be daunting due to its technical complexity, hindering mainstream adoption. Hiero, our AI-powered wallet assistant, alleviates this by allowing natural language interactions for transaction management. Utilizing a Large Language Model trained on Ethereum, Hiero interprets user requests, simulates transactions on a fork for validation, and executes them via interfaces like MetaMask upon approval. This talk will delve into Hiero’s development, showcasing how it simplifies blockchain engagements, ensuring security and accelerating mainstream adoption of DeFi, NFTs, and Web3.



15:15 – 15:30   Zero-Knowledge Machine Learning

Aleksandar Veljkovic
Senior Researcher @ Attic 42, 3327 R&D

In the dynamic realm of artificial intelligence, the fusion of machine learning (ML) and zero-knowledge (ZK) cryptography has birthed the revolutionary paradigm of Zero Knowledge Machine Learning (ZKML). ZKML’s primary objective is to enable verifiable training and inference processes without revealing sensitive data. By leveraging zero-knowledge techniques, ZKML empowers model owners to prove the validity of the predictions against a committed model without disclosing underlying data or model parameters. This addition to the widely adopted ML models may be the key to creating new, decentralized, ML model marketplaces and new business models that utilize the power of ML models in a verifiable environment. The 3327 R&D department at Attic 42 is deeply immersed in advanced research within this domain, aiming to address the challenges of privacy and security in ML applications by developing innovative approaches to solving the underlying challenges of ZKML.



15:30 – 15:45   Cross-chain General Message Passing Protocol

Darko Capko
Professor @ Faculty of technical sciences, University of Novi Sad | CSO and co-founder @ Ethernal

The Cross-Chain General Message Passing Protocol, addressing the increasing need for interconnection among blockchain-based applications is presented. Security challenges inherent to individual blockchains are compounded in cross-chain communication protocols. We analyze the high-level architecture, key components, purposes, and data flow within connected chains, providing an innovative cross-chain solution.



15:45 – 16:15   Panel discussion

Djordje Petrovic & Marko Lazic & Johannes Pfeffer & Aleksandar Veljkovic
Professor @ George Brown college & Blockchain Engineer @ DcentraLab & Founder @ Hiero AI & Senior Researcher @ Attic 42, 3327 R&D
14:30

 —

16:15

Intersection of Data & Blockchain

Djordje Petrovic & Marko Lazic & Johannes Pfeffer & Aleksandar Veljkovic & Darko Capko

MR 12

BS Break

09:30

 —

14:30

BS Break

AI in Public Health

9:30 – 9:45  What We Talk About When We Talk About Real-world Data?

Nevena Veljkovic
CSO @ Heliant

Real-world data (RWD) consist of health information collected from everyday healthcare activities, obtained from sources like electronic health records, insurance claims, and disease registries. The growing ease of RWD access, paired with the rapid progression of artificial intelligence (AI) and machine learning (ML) in healthcare, has heightened interest in their usage. RWD can be inconsistent, often unstructured, and may lack completeness or crucial endpoints due to variability in data entry among providers and health systems. Despite these challenges, RWD presents significant opportunities to address many public health issues that traditional methods may not resolve. This talk will delve into the current applications of RWD and project its future potential in healthcare innovation.



9:45 – 10:00  Unlocking the Potential of Artificial Intelligence (AI) for Rare Diseases

Stefan Zivkovic
Project Coordinator @ NORBS – National Organisation for Rare Diseases of Serbia

The integration of Artificial Intelligence (AI) in healthcare has witnessed substantial growth, showcasing its potential to enhance diagnostic and treatment efficiency. Specifically, machine learning within AI facilitates the analysis of extensive datasets, unveiling patterns and generating crucial insights. In this talk we will explore the transformative impact of AI on addressing the unique challenges associated with rare diseases. We delve into how AI, including machine learning and predictive modeling techniques like deep learning, accelerates the identification of patterns and associations, revolutionizing research and treatment approaches for rare diseases.



10:00 – 10:15  Breaking Communication Barriers

Matija Dodovic
Founder & Machine Learning Engineer @ Sign avatar

Welcome to SignAvatar, the revolutionary sign language software that makes communication easier and more accessible for everyone.By using SignAvatar, businesses and organizations can ensure that they are able to effectively communicate with deaf customers and employees, and improve their inclusivity and accessibility.



10:15 – 10:30  Use of AI in Public Health

Ivana Sesic
Radiology specialist @ University Hospital for Cardiovascular Diseases, Serbia

AI has a significant role in predictive modeling for disease prevention and treatment strategies, and potential to revolutionize healthcare accessibility. In navigating the adoption of AI in public health, a balanced approach to prioritizes ethical principles, data security, and inclusivity is essential to maximize the benefits while addressing the potential challenges. Collaborative efforts between healthcare professionals, policymakers, and technologists are crucial to establishing responsible AI practices in public health



10:30 – 11:00  Panel discussion

Nevena Veljkovic, Stefan Zivkovic, Matija Dodovic, Ivana Sesic
CSO @ Heliant, Project Coordinator @ NORBS – National Organisation for Rare Diseases of Serbia, Founder & Machine Learning Engineer @ Sign avatar, Radiology specialist @ University Hospital for Cardiovascular Diseases, Serbia
09:30

 —

11:00

AI in Public Health

Nevena Veljkovic & Stefan Zivkovic & Matija Dodovic & Ivana Sesic

Breakout Break BS3

11:00

 —

11:30

Breakout Break BS3

People Analytics

11:30 – 12:00  Metrics that matter: Reshaping HR through data

Sara Vatralj
Senior Business Development Manager @ Luppa App

In the absence of a tool on the market that is narrowly focused on employee satisfaction and engagement we didn’t settle for what the market offered—we sculpted our own solution. Founded on the principle that data is the cornerstone of creating exceptional workplaces our solution offers deep analytics. Highlighting compelling metrics, our presentation unveils hidden issues and points to areas demanding innovative HR strategies for an enhanced employee experience.



12:00 – 12:30   Using data to build trust.

Milos Cankovic
CEO and Founder @ Brilliant Buttons

In a landscape where trust in HR is lacking, harnessing the potential of people analytics becomes crucial. Drawing from personal experience in HR, the presentation highlights the disconnect between data collection and its actionable implementation, leading to diminished trust and engagement within organizations. It stresses the significance of continuous feedback for increased engagement and performance, citing Google’s success in using employee feedback for managerial improvements. Moreover, it delves into the shortcomings of periodic performance reviews and advocates for a culture of continuous, open feedback. The introduction of tools like the “”Mindset Meter”” exemplifies the transformative impact of regular feedback sessions in fostering growth and trust within teams. Furthermore, the presentation elucidates the dual role of data, not just for employer benefits but as a tool to support employee development, drawing parallels to how consumer data is utilized by companies like Netflix and Amazon. By leveraging data for personalized development plans, the presentation underscores the transformative potential of empowering employees through data- driven growth initiatives. It concludes by emphasizing the pivotal role of such strategies in retaining top talent and attracting new recruits, advocating for a shift from merely identifying issues to empowering people for continuous development and success.



12:30 – 13:00  Panel discussion

Sara Vatralj, Milos Cankovic
Senior Business Development Manager @ Luppa App, CEO and Founder @ Brilliant Buttons
11:30

 —

13:00

People Analytics

Sara Vatralj & Milos Cankovic

DiplAI – AI for Diplomacy

14:30

 —

17:00

DiplAI – AI for Diplomacy

Branka Panic, Ilija Duni, Kevin LaGrandeur, Monica Nthiga, Masa Njegovan, Tanja Kuzman

CO-LOCATED EVENTS

MR 9

The Dawn of the Social Fintech; AI, Sustainability and Social Science

Keynote
There is a long standing belief that a discipline becomes a science when it can be mathematized. With the advent of AI, the concept of science is widely expanded, with social science being a likely winner. This talk will present a possible view of the future of a largely expanded financial sector drawn by technology and how social science can become a new partner
Keynote
09:15

 —

10:00

The Dawn of the Social Fintech; AI, Sustainability and Social Science

Luis Seco
Professor @ UFT

AI – Digital (r)evolution in banking

Digital transformation of the financial industry
Business talk, Educational talk
Advanced
Digital transformation of the financial industry
10:00