AI engineer

Responsibilities:

• Build and develop a deep AI system: from data and feature engineering to models, online scoring, and monitoring.
• Design anti-fraud features: behavioral, graph-based, temporal, aggregates, velocity checks.
• Write and optimize SQL for data marts, aggregations, and incremental pipelines.
• Develop models: rules → ML → ensembles (GBDT, logistic regression, outlier detection, calibration).
• Implement decision engine (approve / review / reject) with full auditability of decisions.
• Work with event streams and queues (RabbitMQ), ensuring idempotency and reliability.
• Monitor data quality and model performance (PR-AUC, ROC-AUC, recall, latency, drift).
• Perform regular retraining, backtesting, A/B testing, and model improvements.

We expect from you:

• 4+ years of experience.
• Strong Python (pandas, numpy, sklearn; preferably LightGBM/CatBoost).
• Solid SQL skills: window functions, query optimization, incremental processing.
• Understanding of anti-fraud patterns: velocity checks, user/device graphs, anomaly detection, rules + ML approach.
• Experience with production systems: APIs/microservices (FastAPI), queues, logging, monitoring.
• Hands-on experience with imbalanced datasets and threshold tuning based on business cost.
• Understanding of model and business metrics (ROC-AUC, PR-AUC, precision/recall, cost-based metrics).
• Nice to have: MLOps experience (MLflow, Docker, CI/CD), feature pipelines, DWH/ETL tools.