About the Role
Mercury is making a significant, company-wide investment in AI/ML, which is becoming fundamental to various operations including fraud detection, financial crime prevention, credit decisioning, and internal processes. As AI transforms financial services, there's a critical need to redefine model risk management (MRM) for this new era, addressing risks from generative models, autonomous agents, and real-time systems that current frameworks weren't designed to handle. We are looking for a Senior Model Risk Manager - AI/ML who will spearhead this evolution. Ideal candidates may come from traditional model validation with deep experience in modern AI/ML systems, or from model development/applied AI/research with a strong grasp of risk emergence and rigorous challenge in complex systems.
As Senior Model Risk Manager - AI/ML, you will define Mercury’s AI/ML model governance by continuously building and enhancing frameworks. You will own the validation, monitoring, and governance of Mercury’s AI/ML model portfolio, and act as a thought leader in the industry-wide conversation about evolving MRM in the context of AI. You will collaborate closely with data scientists, engineers, compliance leads, and product teams to shape Mercury’s approach and set a new standard for rigorous, forward-looking AI MRM in fintech.
Responsibilities
Model Governance & Monitoring Oversight
- Maintain and enhance Mercury’s model governance framework, including inventory standards, documentation templates, validation standards, and issue management.
- Assess whether first-line monitoring efforts are effective, proportionate to model risk, and sufficient to keep models fit for purpose over time.
Model Validation
- Perform independent validation across predictive ML models, generative AI systems, and agentic workflows, covering data, assumptions, methodology, testing, and monitoring.
- Assess risks in LLM-powered applications, including RAG pipelines, tool use, autonomy boundaries, human oversight, and hallucination risk.
- Identify and document model limitations, failure modes, and emerging AI risks including drift, instability, fairness, and robustness concerns.
MRM Advisory
- Serve as a trusted advisor to data scientists, engineers, product teams, and risk partners throughout the AI/ML lifecycle to provide practical guidance on model risk, governance expectations, and control design without slowing responsible innovation.
- Evaluate new AI use cases for regulatory implications, materiality, and governance requirements prior to deployment.
- Help shape Mercury’s responsible AI standards, including explainability, bias assessment, testing, human oversight, and documentation.
AI Enablement for MRM
- Develop and maintain AI-enabled automation tools to improve the speed, scale, and effectiveness of model governance and validation workflows.
- Modernize the MRM function to operate effectively in a fast-moving AI environment while maintaining strong governance standards.
Culture and Advocacy
- Champion MRM as a strategic enabler of safe and scalable AI/ML adoption, not simply a control function.
- Build model risk literacy across engineering, product, data science, compliance, and risk teams.
Requirements
- Bachelor's degree in a quantitative field (e.g., Computer Science, Engineering, Statistics, Mathematics, etc.) with 6-10 years of meaningful hands-on experience developing or validating AI/ML models and systems, ideally in financial services or fintech.
- Strong technical foundations in Python, SQL, and modern ML tooling (e.g., scikit-learn, XGBoost); familiarity with LLMs, RAG systems, prompt engineering, and AI agent frameworks.
- Experience in evaluating and testing machine learning models (e.g., in fraud detection) and generative AI systems, including custom evals, red-teaming, or frameworks.
- Solid understanding of model risk governance principles and regulatory expectations (e.g., SR 11-7 / OCC 2011-12, SR 26-2).
- Deep appreciation of disciplined model governance and independent effective challenge.
- A healthy dose of skepticism combined with a constructive, solution-oriented approach.
- Comfort operating in ambiguity: capable of synthesizing fragmented technical, operational, and business context into a clear understanding of how complex models and AI systems actually work, and making sound judgments even without a complete playbook or perfect documentation.
- High agency and adaptability: able to operate effectively in a fast-moving environment where priorities evolve quickly, new ad hoc problems emerge regularly, and role boundaries are intentionally broad. You can operate effectively without tightly-defined scope, find the highest-leverage work, and get it done.
- Exceptional attention to detail across documentation, code base, testing artifacts and quantitative analysis.
- Strong written and verbal communication skills; you can explain model risk to a data scientist and to a regulator, and use different language for each.