Research Scientist · ML/DL Engineer
Member of Technical Staff at Krea, working on generative computer vision. Over six years in ML/DL: before Krea I worked at Kandinsky Labs, VILAB, EPFL and the Samsung AI Center, across generative models, computer vision, NLP and meta-learning. I also lecture on a deep learning course and write long-reads about ML, GPU architecture and inference.
View ResumeFeatured Work
Research projects and contributions to the ML/DL community
Open-weights text-to-image foundation model trained from scratch at Krea. The technical report covers data curation, the DiT architecture, the multi-stage training recipe and the style-reference system. Ranked second worldwide on style fidelity.
Open-source LoRA adapters for FLUX.1-dev — realism and a range of art styles — released together with training and inference code. The top adapter reached 500k downloads in its first month on Hugging Face.
Family of text-to-image and text-to-video foundation models. Co-author of the Kandinsky 3.0 technical report and of the Kandinsky 3 framework paper (3.1 is the distilled version, ~20× faster at inference); Kandinsky 4.0 adds video and video-to-audio generation.
Writing
Interactive tools, long-reads, and deep dives into ML, GPU architecture, and inference
Expertise moves from "find the solution" to "build the system that finds it". Oracles, loops, pitfalls — and what remains for the human.
Get in touch
Have an exciting ML research idea, a challenging problem to solve, or just want to connect?