At Ziyatron AI, I developed an EEG-based seizure detection model, fine-tuned it on clinical EEG data, and exported it to ONNX for inference. I also fine-tuned gpt-oss-20b with a custom LoRA for clinical report generation.
At the Image Computing & Machine Learning Lab, Pusan National University, I developed a lightweight EEG-to-fMRI framework that reduced parameters and peak GPU memory while improving reconstruction correlation. This work became a first-author paper at MICCAI 2026.
I also developed EEG architectures evaluated across clinical datasets and a multi-scale CNN–Transformer seizure detection pipeline. My work has led to first-author publications, a granted Kazakhstan utility model patent as a co-inventor, and open-source PyTorch frameworks for EEG research.

