I’m an AI/ML Software Engineer based in Seoul, South Korea, with 6+ years of experience building models across computer vision, LLMs, and multimodal AI — and the software engineering depth to take them all the way to production.

My work sits at the seam between research and engineering — where promising ideas become dependable products. I care most about making ML practical: studying and applying research, designing and fine-tuning models for real use cases, and writing production-quality code so those models actually ship.

Over the years I’ve worked across computer vision, conversational AI, multimodal systems, and production model deployment in both research and product settings — from a university research lab to product teams, and most recently on distributed training and inference systems.

The parts I gravitate toward are model development and the engineering that decides whether it ships: architectures and training workflows, evaluation, inference optimization, and the systems around them. In practice that spans computer vision, LLMs, RAG, and VLMs — backed by solid software engineering across cloud and on-prem environments.

Some of that work is open source — most visibly UniFace, a unified face-analysis library for Python that has grown past 700 stars on GitHub, alongside a family of smaller computer-vision libraries for face parsing, detection, and gaze estimation.

I use this site to write about ML engineering, infrastructure, and applied AI. I also run a YouTube channel where I publish programming tutorials, with a special interest in making technical content more accessible for Uzbek-speaking learners.

Outside of work, I enjoy swimming, hiking, and spending time with friends. I also keep a long-running interest in mathematics and physics, contribute to open source when I can, and occasionally solve algorithm problems on LeetCode to stay sharp on fundamentals.