For a department project, I developed a multi-branch CNN to classify MRI scans into six classes. I tuned its hyperparameters with Optuna and achieved a Macro F1 of 0.9616 ± 0.0151 in stratified five-fold cross-validation.
On a team academic project for a proctoring system, I used a pretrained model to estimate gaze direction by projecting gaze vectors onto the monitor plane. I also contributed to a microservices application launched with Docker Compose.
In my personal project, I fine-tuned OPUS-MT for English-to-Russian translation of slang and colloquial language. I prepared a 12.5k-pair corpus from synthetic translations of imageboard posts and OpenSubtitles dialogue, then evaluated performance with SacreBLEU.
I'm studying Applied Mathematics and Computer Science at Saint Petersburg Electrotechnical University “LETI,” where I’m pursuing a bachelor’s degree. I was a finalist in the 2025 Autonomous Transport Olympiad in the “Autopilots” track.
Russian Federation