ML Developer Offline

What You’ll Be Doing

• Data Exploration & Feature Engineering
  Dive into complex datasets — user activity, financial transactions, KYC logs — and extract meaningful patterns and risk signals
  Design robust and reusable feature sets for training, monitoring, and model explainability

   

• ML Model Developmen
  Build and evaluate models for fraud detection, user risk scoring, payment method ranking, and verification flow prediction
  Balance high class imbalance, cold-start edge cases, and evolving behavioral trends

   

• Segmentation & Profiling
  Develop intelligent user clustering and segmentation frameworks based on multi-dimensional activity and risk signals
  Enable downstream teams to personalize verification, UX, and promo logic
   
• ML Pipeline & Deployment
  Create end-to-end pipelines: data preprocessing → feature stores → training → validation → batch/real-time inference
  Ensure model versioning, reproducibility, monitoring, and drift detection
   
• Process Automation & Risk Tools
  Support automation of high-volume workflows (e.g., withdrawal reviews, user verifications) by integrating risk scores and confidence signals
  Help build internal tooling for analysts and risk teams to interact with model outputs and surface insights

What You BringCore Technical Skills

• Proficiency in Python for data science and ML (e.g., pandas, NumPy, scikit-learn, XGBoost, LightGBM, PyTorch)
• Strong SQL skills for deep-dive investigations and pipeline integrations
• Experience with building production-grade ML pipelines using tools like Airflow, MLflow, Prefect, or similar
• Skilled in feature engineering from raw logs or event streams (both user-level and transactional)
• Solid understanding of model evaluation (especially in high-imbalance, risk-heavy domains)
• Experience with data versioning and managing ML lifecycle in production (e.g., DVC, Feast, or custom solutions)
   

Data & Modeling Mindset

• Strong EDA (exploratory data analysis) skills and the ability to reason about data patterns, anomalies, and edge cases
• Experience in binary classification, ranking, segmentation, and/or semi-supervised detection
• Ability to design models and experiments that reflect real-world operational constraints and decision risks

Bonus Points

• Background in fraud detection, risk scoring, or KYC/AML
• Familiarity with real-time inference pipelines and streaming data processing (Kafka, Flink, Spark Streaming)
• Experience working in iGaming, payments, or financial services
• Exposure to admin or ops tooling — model-driven UIs, analyst dashboards, or labeling workflows