Sanzhar Abdulin

Backend/SQL developer

ALSECO

Maintained production stability of a Java Spring Boot system by investigating misbehaving endpoints, resolving backend logic errors, and applying targeted fixes to ensure data consistency across terabyte-scale PostgreSQL and MongoDB databases • Diagnosed and resolved critical data anomalies by tracing errors through backend code and database state, delivering root-cause analysis and fixes to internal stakeholders. • Automated routine data export and transformation pipelines using Python (pandas, SQLAlchemy) and PostgreSQL stored procedures, reducing daily manual workload by 1.5–2 hours.

Research Assistant

Nazarbayev University

Defended a Thesis project: Lightweight modality-agnostic multimodal HAR model achieving 93.8% accuracy on UTD-MHAD and 98.7% on MM-Fit under strict Leave-One-Subject-Out cross-validation across 4 benchmark datasets. • Developed a Cross-Modal Mamba Fusion Block using bidirectional State Space Models (SSM) for structured cross-modality information exchange with linear O(T) computational complexity. • Applied Recursive Feature Elimination (RFE) with GradientBoosting, reducing feature dimensionality by 50% at zero inference-time cost, improving accuracy by up to 4.2pp on HWU-USP dataset.

Data Science Research Assistant

Central Asian Food Innovation Lab

• Built an end-to-end evaluation pipeline comparing Gemini, LLaMA, OpenAI, and Jenni.AI for automated abstract screening in scientific systematic reviews using Python and NLTK • Configured and maintained multi-sensor camera systems (hyperspectral + depth) for a Central Asian food dataset collection pipeline.

MSc Data Science

Nazarbayev University

BSc Computer Science

Nazarbayev University