Saltanat Zarkhinova

Machine Learning Engineer

Kaspi.kz

Kaspi.kz is a major Kazakhstan-based fintech and e-commerce company, operating the Kaspi.kz super app and marketplace.

Builds and deploys computer vision models and services for Kaspi's marketplace, including product image generation, search, review checks, and inference monitoring.

• Designed and shipped a product background generation pipeline for e-commerce, iterating on model architectures to produce marketplace-ready product photos at scale for production traffic. • Developed and trained a Yolo object detection model for a Search-by-Photo feature, enabling accurate product localization from user-submitted images across the Kaspi marketplace catalog. • Built a product-review image similarity classification model on a frozen vision transformer backbone to automatically verify whether submitted review images match original product listings.

Machine Learning Researcher

Nazarbayev University

Nazarbayev University is a research university in Astana, Kazakhstan, founded in 2010 and known for English-medium teaching.

Researched medical computer vision for skin lesions, including segmentation and detection models; the YOLO26n pipeline reached mAP@0.5 of 0.73.

• Conducted medical AI research focused on interpretable skin cancer classification, developing software pipelines to preprocess and evaluate large-scale image datasets. • Fine-tuned a U-Net segmentation model on the HAM10000 dataset with BCE+Dice loss, contributing to improved IoU on ISIC benchmarks as part of the FSSM-ResNet pipeline. • Fine-tuned YOLO26n for skin lesion detection on a 10k-image dermoscopic dataset, achieving mAP@0.5 of 0.73 within an end-to-end detection and classification pipeline.

Computer Science Major Representative

Nazarbayev University

Nazarbayev University is a research university in Astana, Kazakhstan, founded in 2010 and known for English-medium teaching.

Represented more than 200 computer science students to faculty and organized technical workshops.

• Served as liaison between 200+ students and faculty, improving curriculum feedback channels and organizing technical workshops.

B.Sc., Computer Science

Nazarbayev University