
Risk Management - Compliance Anti-Money Laundering and KYC - Model Risk Program Associate
- Jersey City, NJ
- Permanent
- Full-time
- Perform model reviews: analyze the conceptual soundness of compliance model and assess model behavior and suitability in the context of usage.
- Guide on model usage and act as the first point of contact for the business on all new models and changes to existing models.
- Develop and implement alternative model benchmarks and compare the outcome of various models. Design model performance metrics.
- Liaise with model developers, users, and compliance groups, and provide guidance on model risk.
- Evaluate model performance on a regular basis.
- 1 plus years of experience in a quantitative modeling role, such as Data Science, Quantitative Model Development, Model Validation, or Technology focused on Data Science.
- A PhD or Master's degree in a quantitative field such as Mathematics, Physics, Engineering, Computer Science, Economics or Finance is required.
- Strong verbal and written communication skills, with the ability to interface with other functional areas in the firm on model-related issues and write high quality technical reports.
- Deep understanding of standard statistical techniques, such as regression analysis.
- Hands-on experience with standard Machine Learning models, including Boosted Trees, Neural Networks, SVM, and LLM (e.g. BERT).
- Experience of working with dedicated ML packages, such as TensorFlow or similar, as well as data processing and numeric programming tools (NumPy, SciPy, Pandas, etc.).
- Ability to implement benchmarking models in Python, R, or equivalent.
- Risk- and control-oriented mindset: ability to ask incisive questions, assess the materiality of model issues, and escalate issues appropriately.
- Ability to work in a fast-paced, results-driven environment.
- Prior experience in modeling, reviewing or managing models for sanctions screening, trade surveillance or transaction monitoring is desirable.
- Experience with database interfacing, data management, and preprocessing (e.g. SQL or kdb+, q) is a plus.