Kuat Zharykbayev

Machine Learning Engineer (Apr. 2025–Present); previously Machine Learning Engineer / Researcher (Feb. 2025–Apr. 2025)

SpaceLab LTD

Built NLP, OCR, and animal-monitoring computer-vision systems, and a scientific data pipeline for testing solar–seismic associations.

• Built an end-to-end Russian-language NLP pipeline over 125,894 Telegram messages (86,316 after dedup and URL-only removal), fine-tuning RuBERT into a 3-class behavioral-text classifier: 0.84 macro-F1 on a lexicon-stratified eval set, random held-out macro-F1 on a random held-out sample. Labels were bootstrapped through an LLM-assisted pipeline and audited by blind human re-annotation (Cohen’s κ = 0.81); scored the full cleaned corpus for downstream event analysis.

Research Assistant

Kazakh-British Technical University (KBTU)

A university in Kazakhstan.

Contributed to research on perceptual color extraction and retrospective seismicity analysis.

• Co-authored research on human-perception-based dominant color extraction, currently under review at a Q1 Elsevier journal (see Publications). • Contributed to research of retrospective seismicity analysis of the Zailiysky Alatau region, integrating solar-geomagnetic and geochemical precursor data.

Master of Science, Data Science / Scientific Research

Kazakh-British Technical University (KBTU)

Bachelor’s Degree, Computer Science / Machine Learning / Robotics

Kazakh-British Technical University (KBTU)

Exchange Program, Artificial Intelligence / Machine Learning

Gwangju Institute of Science and Technology (GIST)