Marko Stefanović

Led end-to-end development of production-ready Generative AI products combining React frontends, LLM inference, and vector memory systems. Designed LLM agents featuring tool use, memory, and chain-of-thought reasoning for document Q&A and intelligent assistants. Integrated semantic search technologies such as FAISS and pgvector with LangChain pipelines for RAG-based user experiences. Directed multi-agent architecture including stateful LLM flows using LangGraph. Fostered cross-functional collaboration with designers, product owners, and backend teams.

Led end-to-end development of production-ready Generative AI products combining React frontends, LLM inference, and vector memory systems. Designed LLM agents with tool use, memory, and chain-of-thought reasoning for document Q&A and intelligent assistants. Integrated semantic search (FAISS, pgvector) with LangChain pipelines for RAG-based user experiences. Directed multi-agent architecture, including stateful LLM flows using LangGraph. Drove cross-functional collaboration with designers, product owners, and backend teams.

Built and optimized Retrieval-Augmented Generation (RAG) pipelines for LLM applications serving healthcare, agriculture, and enterprise data analysis sectors. Developed LangGraph-based multi-agent systems and integrated OpenAI GPT APIs with various third-party services. Deployed scalable, latency-optimized solutions on AWS. Contributed to internal tooling for ETL processing, embedding management, and pipeline monitoring.