DSC Europe 24 SCHEDULE

Description:   Redefining Business with AI will explore the transformative role of artificial intelligence in reshaping industries and business models. This session will focus on how AI-driven innovations are optimizing operations, enhancing decision-making, and unlocking new revenue streams across sectors. Attendees will gain insights into practical AI applications, from automating routine tasks to creating personalized customer experiences, and discover how businesses are leveraging AI to stay competitive in an increasingly data-driven world.

---

10:00 – 10:15   Redefining Business with AI

Tervel Sopov
Partner & AI Strategy and Data Science Lead for Central Europe @ Deloitte

It has been two years since GenAI made its public launch. According to our survey two thirds of companies are increasing their investments in AI as they see strong value to date. How is AI transforming the business? What are the challenges companies face? And which strategic consideration should you take into account for successful adoption? Join us for an interactive session with AI&Data leaders.

---

10:15 – 10:30   Bridging departments with AI: Real-World Use Cases in Telecom’s Key Functions

Sinisa Arsic
Director of Department for Data Analytics and Intelligent Automation of Business Processes @ Telekom Srbija

TBA

---

10:30 – 10:45   Redefining Business with AI

Markus Warg
Sales Manager, EMEA @ Amber

TBA

---

10:45 – 11:00   Scaling AI beyond pilots: strategies for widespread implementation

Denitsa Gavrilova
Director AI and Data @ Yettel.Hungary

TBA

---

11:00 – 11:15   TBA

Vladimir Cuk
Head of Marketing and Ecommerce, Analytics @ Gigatron

TBA

---

11:00 – 12:00   Panel duscussion

Moderator: Tervel Sopov
Partner & AI Strategy and Data Science Lead for Central Europe @ Deloitte

Description:   Leveraging AI for Customer-Centric Strategies focuses on how artificial intelligence is transforming the way businesses understand and engage with their customers. This session will explore innovative methods such as data psychology to uncover behavioral insights and AI-driven tools that empower non-technical business users. From scaling empathy through data-driven nudging techniques to leveraging AI platforms for personalized marketing and customer engagement, this session provides actionable strategies for businesses looking to drive growth and enhance customer satisfaction through AI-powered solutions.

---

12:00 – 12:15   How To Use Data Psychology to Scale Empathy and Drive Sales

Dr Vanja Ljevar
Co-Founder and Chief Data Scientist @ KUBIK Intelligence

Every online action leaves a digital trace, revealing more about us than we realise. Our digital footprints hold clues to our psychology and habits, forming a puzzle that, when pieced together, reveals a comprehensive picture of our inner lives. At Kubik we base our approach on innovative data science techniques that empower us to intelligently extract data-derived behavioural insights. This serves as a basis for our psychology-based nudging techniques that resonate most with customers and create your ultimate competitive edge. By combing data science and behavioural science, we aim to uncover an individual’s lifestyle, personality, and behaviour, effectively revealing their “digital DNA.”

---

12:15 – 12:30   Empowering Business Users with AI: Unlocking the Power of Data with Solver Suite

Nikola Trajkov
Machine Learning Engineer – Team Lead @ Things Solver

In this talk, we’ll introduce you to Solver Suite, an AI-powered platform designed to democratize data and empower business users across marketing, sales, and customer service departments. With Solver Suite, businesses can leverage advanced machine learning models and data pipelines without the need for large upfront investments or specialized teams. Through an intuitive interface, users can explore, segment, and personalize data, driving impactful customer engagements. Whether it’s through automated marketing flows, personalized content placement, or targeted communication, Solver Suite helps businesses transform raw data into actionable insights that fuel growth and customer satisfaction.

---

12:30 – 12:45   Transforming Brand Reputation Management through AI-Driven Social Media Monitoring and Analysis

Natalia Drazhnik
Product owner @ YCH Blue Digital Limited

In today’s fast-paced digital environment, effective social media monitoring and analysis is essential for managing brand reputation and shaping strategic communication. My presentation will explore how AI-powered tools can overcome the limitations of traditional monitoring, automating data collection, enhancing sentiment analysis, and providing real-time insights. We’ll examine the unique challenges posed by high volumes of social media interactions—such as identifying reputational risks and addressing misinformation. I’ll share examples of how AI-driven solutions can detect early signs of crises, support proactive engagement, and enable data-driven decisions that build customer trust. For organisations aiming to improve their reputation management and communication strategy, AI offers a powerful, scalable solution to stay responsive to audience needs and maintain a positive brand image.

---

12:45 – 13:00   Scaling Customer Support with AI: LLM-Powered Email Management for Modern Call Center

Andjela Mihajlovic
AI Engineer @ Comtrade System Integration

By leveraging Azure OpenAI’s advanced language models, we redefined how customer communication is managed. The solution categorized emails, identified customer request types, and analyzed sentiments to help the company proactively reshape offerings and enhance retention strategies. Our support team can now focus on what they do best: solving complex inquiries that require deep expertise, while AI handles the routine issue, saving them valuable time.

---

13:00 – 14:00   Panel duscussion

Moderator: Walid G. H. Saad

Description:   It is well known that language models perform much worse for languages other than English or a few other major global languages supported by large linguistic communities and strong economies. South Slavic languages are considered low-resource languages. In this session, we will meet individuals and teams who have taken up the challenge and decided to create something despite resource limits in every way – from data and GPU resources to financial means and time. At the same time, they make their work available to a wider community under permissive licenses, actively contributing to our collective progress and allowing us to fully realize the potentials of the new generation of artificial intelligence in our native language.

Moderator: Slobodan Markovic
Digital Advisor @ UNDP

---

15:00 – 15:15   Presentation

Nikola Trajkov
Machine Learning Engineer – Team Lead @ Things Solver

Nikola will talk about his decision to develop the first large language model for the Macedonian language, the challenges he faced along the way, and how he overcame them. He will talk about community feedback, discuss potential applications of the model, highlight areas for improvement, and suggest where the community should focus its efforts to enhance the position of Macedonian (and other regional low-resource languages) in the age of AI.

---

15:15 – 15:30   Presentation

Mihailo Skoric
Research Associate and Project Leader @ Faculty of Mining and Geology, University of Belgrade

Mihailo will present the work he has done as part of the TESLA project (Text Embeddings – Serbian Language Applications). This includes cleaning and consolidating the largest regional pretraining web and scientific corpora, which cover content in Serbian, Croatian, Bosnian, and Montenegrin. He will also showcase several models trained on the resources prepared by him and his colleagues from the Serbian Association for Language Resources and Technologies (JeRTeh). Additionally, he will walk us through the challenges encountered during this work, identify areas for further improvements, and share what’s next for him and his colleagues.

---

15:30 – 15:45   Presentation

Vuk Batanovic
Research Assistant @ Innovation Center, School of Electrical Engineering, University of Belgrade (ICEF)

Vuk will present the results of the COMtext.SR project – the first corpus of Serbian legal texts (in both Ekavian and Ijekavian variants) manually labeled for morphosyntactic tagging, lemmatization, and named entity recognition, as well as large language models fine-tuned for these tasks in the legal domain. He will discuss the challenges faced in the development of these resources and tools, and the opportunities for using them in various practical applications. Vuk will also talk about other corpora and models developed for South Slavic languages by the NLP lab at the Innovation Center of the School of Electrical Engineering in Belgrade, as well as those created by the Regional Linguistic Data Initiative (ReLDI), an association focused on the development of publicly available resources, tools, and benchmarks for natural language processing.

---

15:45 – 16:00   Presentation

Teodora Mihajlov & Selena Milutin
Machine Learning Intern @ SmartCat & Machine Learning Intern @ SmartCat

Embedding models are foundational to modern NLP applications, yet their development and training demand substantial data and computational resources. As a result, most available models are tailored to larger languages, predominantly English. In this talk, Teodora and Selena will present their 2024 SmartCat internship project, focused on creating an information retrieval embedding model specifically for the Serbian language. They will discuss the motivation behind this project, share insights into dataset preparation, and describe their approach to model training and evaluation. The talk will also explore potential applications of the model and areas for future improvement.

---

16:00 – 16:15   Presentation

Andrija Sagic
Head of Digital Development Department @ Library “Milutin Bojic”​

Andrija will discuss his work on fine-tuning the OpenAI Whisper model for Automated Speech Recognition (ASR) in the Serbian language. He will describe the process of cleaning, consolidating, and creating new fine-tuning datasets. Additionally, he will share how this model is supporting his daily work and provide his perspective on what is needed to advance Serbian language technologies from the viewpoint of a cultural heritage professional.

---

16:15 – 16:30   Presentation

Dragan Jovanovic
CTO & Iskon.ai​

Dragan from Iskon.AI will present his work on continued pre-training of a large language model (Mistral-7b) to understand and generate text in new languages, including Serbian, Bosnian and Croatian. Rather than depending only on extensive datasets in the target language, their method utilizes a more compact set of both synthetic and human-curated data.