Amir Garifulla

Senior Data Scientist

Freedom Bank

Kazakhstan-based commercial bank.

Improved a risk model's ROC-AUC from 73% to 76% and validated gains through controlled experiments.

• Improved ROC-AUC of a gradient-boosted risk model from 73% to 76% through systematic feature engineering, feature stability (PSI) filtering, adversarial validation, and Bayesian hyperparameter tuning with OPTUNA • Ran controlled model experiments across multiple data sources, applying rigorous evaluation methodology to validate performance gains before deployment

Middle Data Scientist; Junior Data Scientist

Kaspi Bank

Kazakhstan-based bank within the Kaspi ecosystem.

Progressed from Junior to Middle Data Scientist, improving fraud recall, approval rate, and risk outcomes while owning a risk-scoring pipeline.

• Reduced model-driven risk exposure from 4.8% to 2.8% through data-driven threshold tuning and rule optimization, balancing precision/recall trade-offs at scale • Owned end-to-end model lifecycle for a new product’s risk-scoring pipeline: model selection, threshold calibration, KPI design, and post-deployment monitoring • Redesigned anomaly/fraud detection rules and decision thresholds, tripling recall from 20% to 60% while maintaining precision targets

Master, Data Science

Kazakh-British Technical University

Bachelor, Information and Communication Technologies

Suleyman Demirel University