Daniel Bartolic

AI Image Generation Expert & Full-Stack Developer

22 · Ljubljana / Koper, Slovenia

I specialize in deploying open-source AI models to production. I can take any open-source model available and deploy it to a serverless API — with deep hands-on experience in models like Qwen Image 2512, Flux, Z-Image, Wan 2.2 for image and video generation, and Qwen 3.5 for custom LLM chatbots. I build end-to-end pipelines from model weights to production endpoints using ComfyUI, vLLM, and llama.cpp — and optimize results by combining multiple endpoints to get the best quality at the lowest cost.

With a background in pure mathematics from the University of Ljubljana, I moved into AI full-time — and am currently building realstagram.ai, a SaaS platform for AI image generation.

Every character I create is a custom-trained LoRA model. I design end-to-end pipelines from model training through ComfyUI workflows to serverless deployment on RunPod and Vast.ai. I also run an AI Instagram account showcasing my workflow outputs.