Lead Data Scientist

Lead Data Scientist

Requirements

• Graduate degree in CS, Data Science, Statistics, Machine Learning, AI or a related STEM field.
• Demonstrated ability to independently contribute to overall team objectives.
• Strong background in statistics, probability, and linear algebra as applied to machine learning models.
• Data science and data engineering experience.
• Proven experience with supervised and unsupervised learning techniques, such as multiple of XGBoost/LightGBM/GBM, deep neural networks, Isolation Forest, and clustering.
• Strong grasp over data science and machine learning concepts.
• Experience with SQL language and one or multiple of PySpark, Hadoop, Impala, Hive.
• Good knowledge of Linux / Bash environment.
• Python, Pyspark.
• Knowledge of model optimization techniques.
• The ability to work closely with more senior data scientists implementing and optimizing models within big data pipelines.
• Good communication skills.
• Highly skilled problem solver.
• Exhibits a high degree of initiative.

Responsibilities

• Work closely with the business owners to understand business requirements, performance metrics regarding data quality and model performance of customer facing products.
• Lead the development of advancing fraud detection models specifically credit and debit card transaction level models.
• Lead the efforts to enhance the best modeling practices that maintains the competitiveness of our fraud detection models.
• Oversee implementation of data and model development pipelines.
• Explore fraudulent patterns or trends for feature discovery and enhance fraud detection model performance.
• Manage the testing of trained models to ensure their robustness and assess their readiness for deployment.

Preferred Qualifications

• Experience building payment fraud detection models.
• PhD in CS, Statistics, or a related quantitative STEM field.
• Experience with data engineering in PySpark on petabyte scale data.
• Expertise in timeseries analysis and forecasting techniques, such as ARIMA, RNN, and LSTM networks to detect anomalies in large-scale, time-sensitive datasets.
• Experience with active learning methods, particularly in situations where labeled data is scarce or expensive to obtain.
• Understanding of data privacy, algorithm bias, and developing fair, transparent, and accountable algorithms.
• Understands and implements methods to evaluate own work and others for error.
• Loves working with error-prone, messy, disparate, unstructured data.