Role Summary
BMO’s Enterprise Risk group is seeking a hands‑on Credit Risk Analyst to develop, calibrate, deploy, and monitor regulatory Probability of Default (PD) models for wholesale portfolios. With a primary emphasis on data science and model development (60%) and complementary ownership of production engineering and MLOps (40%), you will translate complex business requirements and regulatory expectations into robust, explainable, and auditable solutions. The ideal candidate blends quantitative rigor with strong comprehension, communication, critical thinking, and collaboration skills.
Key Responsibilities
A) Data Science & Modelling - ~60%
- Own the end‑to‑end lifecycle of wholesale PD models (scoping, data understanding, feature engineering, model development, calibration, performance measurement, documentation, implementation support, and ongoing monitoring).
- Design PD modelling strategies across wholesale segments (e.g., corporate, commercial, financial institutions, specialized lending), selecting appropriate techniques (e.g., logistic/GLM, survival/PD term structures, regularization) with clear rationale and interpretability.
- Engineer features from internal/external data (financial statements, rating histories, obligor/industry attributes, collateral/covenants, macroeconomic drivers) ensuring lineage, quality, and reproducibility.
- Establish and track discrimination, calibration, and stability metrics; perform back‑testing and benchmarking; develop champion‑challenger frameworks and sensitivity/stress analyses.
Produce high‑quality model documentation tailored to model validation, internal audit, and regulators; communicate complex concepts clearly to technical and non‑technical stakeholders.
B) Machine Learning Engineering & Delivery - ~40%
- Develop models on AWS (e.g., SageMaker, S3, Glue, EMR/Athena, Lambda/Step Functions) with CI/CD, version control, and automated testing (unit, integration, UAT).
- Package and expose scoring services and monitoring jobs (e.g., Python APIs with FastAPI/Flask; batch pipelines) following secure‑by‑design patterns and enterprise standards.
- Implement data pipelines and model monitoring (data drift, performance drift, stability, alerts) with robust logging, lineage, and access controls.
- Partner with Data Engineering and Technology to align on architectures (containerization with Docker; orchestration; model registry) and ensure reliability and scalability.
Domain Focus: Regulatory Wholesale Credit Risk
- Develop PD models aligned to wholesale credit risk regulations (stress testing, allowance).
- Support regulatory and accounting use cases (e.g., CCAR - DFAST stress testing, IFRS 9 provisioning, CECL provisioning) and model risk management expectations across the model lifecycle.
- Collaborate closely with stakeholders across Risk, Lines of Business, Model Validation, Internal Audit, and Compliance during reviews, findings remediation, and regulatory exams.
Required Qualifications
- 3+ years of hands‑on experience building and implementing credit risk models—ideally wholesale PD—with measurable business impact.
- Advanced proficiency in Python (e.g., pandas, NumPy, scikit‑learn, statsmodels, XGBoost/lightGBM as appropriate).
- Working proficiency in SAS for data preparation, analytics, and/or model implementation within enterprise environments.
- Practical experience on AWS (e.g., S3, Glue, EMR/Athena, SageMaker, Lambda/Step Functions) for data processing and model deployment.
- Strong understanding of statistical learning, feature engineering, validation techniques, and performance monitoring for PD models.
- Proven ability to author clear, regulator‑ready documentation and to present complex analyses to senior stakeholders.
- Bachelor’s degree in a quantitative field (Statistics, Mathematics, Computer Science, Engineering, Economics/Finance); graduate degree preferred.
Preferred / Nice to Have
- Experience with wholesale data domains (financial spreading, internal ratings, default events, collateral/covenants, industry taxonomy) and linking to macroeconomic variables.
- Exposure to MLOps tooling (e.g., MLflow/Feature Store, Docker/Kubernetes, Git, CI/CD) and workflow orchestration (e.g., Airflow/Step Functions).
- Working knowledge of SQL and distributed data processing (e.g., Spark); experience with SAS macros is an asset.
- Familiarity with model risk management practices and regulatory expectations across the lifecycle; experience engaging model validation, audit, and regulators.
Meta Skills & Ways of Working (Core to Success)
- Comprehension: rapidly absorb complex business and regulatory context; ask incisive questions to frame the real problem.
- Communication: tailor messages for technical peers, senior executives, validators, and regulators; write clear, concise documentation.
- Critical Thinking: evaluate trade-offs between interpretability and performance; design tests to falsify assumptions; identify and mitigate model risk.
- Colla