Antifraud Analyst

We are looking for a candidate for a cross-functional role that combines tasks related to online fraud and cyber threat protection, as well as system and infrastructure security.

What needs to be done:

1. ML modeling for antifraud: Development, training, and implementation of machine learning models for fraud detection in transactional and session traffic – classifiers (XGBoost, LightGBM, neural networks), anomaly models (Isolation Forest, Autoencoder), behavioral pattern clustering.
2. Scoring systems: Design and support of ML scoring – behavioral, transactional, identification scoring; calibration of probability estimates, model drift monitoring.
3. Feature Engineering: Extraction and construction of features from raw session logs, transactions, device fingerprints, and graph connections between entities.
4. Traffic analysis and anomaly detection: Analysis of session and transactional traffic to identify fraud patterns using statistical methods and unsupervised ML – outliers, atypical action sequences, behavioral anomalies.
5. Model validation and testing: Testing and offline validation of antifraud models: quality metrics (AUC-ROC, KS, Gini, F1), error analysis, False Positive / False Negative control.
6. Rule management and hybrid systems: Integration of ML models with a rule-based engine: trigger setup, backtesting, threshold calibration, unified decision-making pipeline support.
7. Mitigation of automated threats: Application of ML for detecting bot activity, multi-accounting, bonus hunting, scraping.
8. Monitoring and incidents: Development of ML metrics and dashboards for online model quality monitoring; participation in fraud incident investigations, model retraining when attack patterns change.

What we expect from the candidate:

1. Practical experience in building ML models for fraud detection or related tasks (credit scoring, anomaly detection, churn prediction).
2. Proficient in Python and the ML stack: scikit-learn, XGBoost/LightGBM, optionally PyTorch/TensorFlow for sequence models.
3. Experience working with large volumes of unstructured data: raw logs, event streams, transaction chains.
4. Confident SQL for working with analytical data warehouses.
5. Understanding of binary classification quality metrics and business impact assessment of models.
6. Bonus points for: Experience with graph methods for detecting linked fraudulent networks (Graph Neural Networks, community detection); working with streaming data (Kafka, Flink) for real-time online scoring; integration of external scoring APIs and data providers.

Why it's good to work with us:

1. Personalized salary offer and official employment, we believe in transparency and honesty.
2. Work schedule: 5/2. Office-based work in Yekaterinburg.
3. Voluntary medical insurance with dental coverage, because we care about your health.
4. Career consultant assistance for choosing an individual development path.
5. Work in an accredited IT company.
6. Headquarters in the center of Yekaterinburg with its own bar, relaxation areas, amphitheater, and a garden of date palms, ferns, and giant ficus trees.
7. Partial meal compensation for office work.
8. Open dialogue with top managers and a friendly team – the foundation of our culture.
9. English language study, compensation for training, and participation in conferences to help you grow with the company.
10. Corporate discount at Zolotoye Yabloko, plus bonuses and promotions from partners.
11. Large-scale corporate events, book and sports clubs, small talk, and many other informal gatherings.
12. No dress code or appearance restrictions. We respect your individuality!