Exploring LLMs: Insights from AI Open Day in Skopje

Large language models (LLMs) — an advanced type of artificial intelligence (AI) that processes and generates human-like text — are quietly becoming a necessary tool for efficiency and cost-savings in every industry.?

Whether the task is summarizing patient data, generating software code, or powering?customer support chatbots, LLMs and AI are working in the background to improve how humans interact with machines through language.?

Last month, HTEC hosted AI Open Day, a conference in Skopje, North Macedonia focused on the advantages and growing pains of LLMs. ?

Through a series of sessions, HTEC experts — including senior data scientists Ognjen Milicevic , Jelena Banjac Luki? , Marko Karbevski , and senior software engineer, Dimitar Iliev ?? — explored various LLMs and AI topics. Session subjects included how to build cloud applications with integrated custom copilots and how to use LLMs to analyze large text datasets to detect linguistic trends.?

Read on to learn more about each of the AI Open Day sessions and access conference insights from Ognjen Mili?evi?, and conference organizer, Hristina Panovska , head of health & life sciences technology at HTEC. ?

LLMs and AI: tech innovation meets everyday use?

While the sessions at AI Open Day focused on highly technical aspects of LLMs, HTEC’s Ognjen Mili?evi? emphasizes that successful LLM adoption combines technical innovation with the practical use of generative AI (GenAI) by non-technical workers. The use of LLM-powered tools in the workplace is growing rapidly, with?37% of US professionals using AI chatbots every day in their work, according to a recent LLM survey by Amperly.?

Non-technical AI skills worth honing include learning how to craft clear prompts for AI systems, evaluating the accuracy of AI-generated responses, and confidently navigating new AI features in common workplace tools like Microsoft Office and Google Workspace.?

Proper prompting can help reduce AI hallucinations or inaccurate responses. Download our white paper, “When machines dream: Overcoming the challenges of AI hallucinations”, to learn more.?

“As companies integrate LLM tools into workflows, individuals must improve their AI competency — understanding its capabilities, refining prompts, and interpreting outputs.” ? — Ognjen Mili?evi?, senior data scientist at HTEC??

While non-technical workers develop new skills to work with LLM-based tools and increase AI competency, software developers and data scientists must also adapt to the changing technological landscape to capitalize on the potential of LLMs. ?

With this in mind, session speakers offered up practical guidance for data scientists, developers, and product teams on LLM design, construction, features, and applications. ?

AI Open Day session recaps??

Here are brief summaries of the four conference sessions led by HTEC data scientists:?

LLM application for standardization of biological databases?

Speaker: Ognjen Mili?evi?

In this session, Ognjen focused on AI usage in life sciences, exploring how to use LLMs and AI to unify complex biological databases and improve data accessibility and accuracy, which are critical for advancing research in the life sciences field.?

Click here to watch Ognjen’s entire AI Open Day session?

Building custom copilots: because even superheroes need sidekicks?

Speaker: Dimitar Iliev?

This session provided an introduction to OpenAI and GPT models, followed by an overview of Azure OpenAI and a sample copilot architecture. Dimitar also explored the three components of an AI agent: plugin, planner, and persona. Finally, the session wrapped up with best practices for securing intelligent applications in the cloud and key considerations when building AI copilots.?

Click here to watch Dimitar’s entire AI Open Day session??

Possible modifications to the attention mechanism?

Speaker: Marko Karbevski?

During this session, Marko explored transformer architecture — a neural network design that uses a "self-attention" mechanism to help LLMs process and generate text effectively. The session focused on how to reduce parameters and boost performance in an optimized LLM architecture.? ?

Click here to watch Marko’s entire AI Open Day session?

?Applied LLMs: trend detection in scientific papers?

Speaker: Jelena Banjac Luki??

In this session, Jelena demonstrated how to use LLMs to detect trends in the use of Serbian words with English roots in technical scientific papers published by the University of Novi Sad from 2008-2024.? ?

Click here to watch Jelena’s entire AI Open Day session?

LLM growth is a community effort?

HTEC’s Hristina Panovska said a major highlight of the conference was the opportunity to connect with professionals from diverse fields.??

“Engaging with individuals who approach problems from different perspectives is at the core of how we generate new ideas, and?it underscores the versatility of LLMs,” Hristina said.?

She added that the conference also demonstrated the importance of establishing tech communities to give AI innovators a place to share experiences, learn, and solve problems together.??

Summing up the AI Open Day conference in Skopje, Hristina?said:?

“We brought experts together from diverse disciplines to showcase the wide-ranging applications of LLMs, from cutting-edge research to real-world problem solving. Our goal was to share insights on the AI space, and we look forward to continuing this series and contributing to collective knowledge sharing in the field.”?

? For enterprises looking to broaden their AI usage, HTEC is actively helping clients pilot and deploy production-ready AI solutions across their organizations ethically and securely.??

Are you ready to drive the future of artificial intelligence and large language models? Connect with an HTEC expert.?