Elvedin Dobrnjic

At Ikon Srl, I researched generative AI APIs and tested locally run open-source LLMs. I built Python middleware to orchestrate and integrate multiple models, with dynamic model loading management.

For my thesis at the University of Udine, I developed a unified framework for dynamic orchestration of generative AI models in hybrid environments via custom APIs. I also built a personal RAG pipeline that stores document embeddings in PostgreSQL with pgvector and uses semantic search to retrieve context for an LLM.

In a personal data analysis and ML project, I analyzed benchmark results from multiple LLMs and APIs, then trained and evaluated baseline models with scikit-learn. I’m looking for an opportunity in junior or full stack development, data analysis or machine learning, or AI integration.