Engineering with LLMs: Knowledge Graphs and Feature Development

Otvaramo 2026. sezonu Heapspace događaja sa dva praktična predavanja o LLM-ovima i modernom razvoju softvera:

• Knowledge Graphs Meet Generative AI – Marko Putniković, Software Engineer @Bosch
• Feature Development in the Age of LLMs – Jovana Kesler, Front End Engineer @Decenter

Ova predavanja pokrivaju praktičnu stranu LLM-ova u softverskom inženjeringu. Naučićete kako ih koristiti u ozbiljnim sistemima, kako održati kontrolu i strukturu, i kako prelaziti sa eksperimenata na produkciju, uz praktične savete kako raditi sa AI-om na način koji ostavlja kod čistim i održivim.

Događaj se održava u Vidikovac Beograđanka Coworking Space u četvrtak 5. februara od 18.30h.

Potvrdi dolazak ovde.

***

Marko Putniković, Software Engineer @Bosch
Knowledge Graphs Meet Generative AI

Knowledge Graphs Meet Generative AI: during this presentation, we will introduce knowledge graphs (KGs) and the core technologies used to build and maintain them. We will then explore the synergetic relationship between KGs and large language models (LLMs), explaining how one can enhance and benefit the other. We will also present current examples of how Bosch has leveraged LLMs to construct large-scale KGs, and conclude with some open ideas on how this LLM-KG synergy could be further utilized within Bosch.

Marko has been working at Bosch as a Software Engineer (in AI) since 2022. He's a PhD candidate at the University of Belgrade. His main work and research areas are Software Engineering and Natural Language Processing.

Jovana Kesler, Front End Engineer @Decenter
Feature Development in the Age of LLMs

The role of the Software engineer has undergone a radical shift. We are moving beyond traditional implementation into a role of AI wranglers, constantly adapting our 'prompting language' to extract the most efficient, production-ready results.​In this talk, we’ll dive into the modern development loop, focusing on the coding phase, more specifically from the front-end perspective. We will tackle the high-stakes challenges emerging from this transition: maintaining architectural integrity when code is generated at scale, and ensuring that today’s 'AI velocity' doesn’t become tomorrow’s unmaintainable technical debt. This is a practical session for leads and those who identify as such, on how to ship faster while keeping the codebase clean and robust.

***

Heapspace Meetups okuplja radoznale ljubitelje tehnologije iz različitih oblasti IT-a. Kroz organizovanje događaja edukativnog karaktera, želimo da podstaknemo neformalnu edukaciju iz ovih oblasti i time pružamo zajednici prostor za dalje umrežavanje i učenje.

Vidimo se!