Senior Data Scientist, AI & Model Risk

As a Senior Data Scientist focusing on AI & Model Risk, you will lead and coordinate AI risk assessments for Generative AI and Large Language Model use cases, applying Model Risk Management principles to ensure safe, compliant, and responsible deployment. This role sits at the intersection of Model Risk Management and AI Governance, with close coordination with Information Security, Compliance, and Legal stakeholders.

You will tackle complex and ambiguous challenges, applying sound judgment to select appropriate methodologies and drive risk-based decisions with minimal day‑to‑day oversight. You will partner with senior stakeholders across Risk, Engineering, Legal, Compliance, and Information Security to drive assessment outcomes and continuously strengthen governance practice. This role places you at the center of an innovative team within SFS that is actively shaping how AI risk management works in practice, operationalizing principles into processes that enable safe, effective AI deployment.

If you’re excited by the challenge of shaping how AI gets deployed safely and responsibly in financial services, this is the place to be.

• Lead end‑to‑end AI Risk Assessments for generative AI and LLM use cases across the Bank; embed in Block’s enterprise‑wide GenAI review process, coordinate cross‑functional SMEs (Legal, Compliance, Info Sec, Data Governance, MRM, ERM, BRC, TPRM, Financial Crimes), and manage timelines to ensure reviews are completed within SLA.
• Review AI system design and documentation; including retrieval sources, assumptions, limitations, fallback plans, guardrail configurations, and change management procedures — across banking use cases such as fraud detection, BSA/AML compliance, credit decisioning, and customer‑facing applications, ensuring governance controls are commensurate with each use case’s risk profile.
• Assess pre‑deployment testing for adequacy inclusive of output integrity, hallucination detection, boundary and edge case testing, ethical and safety guardrails, bias testing, A/B testing, volume testing, and UAT — designing and conducting independent testing as needed.
• Evaluate ongoing monitoring plans for comprehensiveness — including accuracy, hallucination rates, drift detection, sensitive data controls, reliability metrics, CSAT, acceptable performance ranges, and documented remediation procedures.
• Develop and maintain templates, tools, and procedures to support the effectiveness and scalability of the AI Risk Governance Program.
• Monitor the evolving regulatory landscape for AI in banking — including FFIEC IT Examination Handbook standards, FDIC Financial Institution Letters, interagency statements, and the anticipated RFI on AI model risk management referenced in SR 26-2 — and incorporate emerging guidance into the AI risk governance program; support SFS’s response to regulatory inquiries as needed.

• A minimum of 5 years of related experience with a Bachelor’s degree in a quantitative field; or 3 years and a Master’s degree; or a PhD without experience; or equivalent work experience in risk management, model risk management, or AI risk management.
• Proficiency in Python or similar languages for evaluating AI system behavior, writing test scripts, or analyzing model outputs.
• Strong understanding of generative AI architectures; including LLMs, transformer models, RAG systems, and agentic AI, plus hands‑on experience interacting with and critically evaluating these systems, sufficient to assess design decisions, output quality, and limitations.
• Understanding of interagency model risk management principles, including SR 26-2.
• Knowledge of AI testing methodologies, e.g. functional testing, bias testing, adversarial testing, and performance monitoring plus familiarity with data privacy and security principles (encryption, access controls, data classification).
• Excellent written and verbal communication and the ability to translate complex technical AI concepts for non‑technical stakeholders, senior management, and regulators.
• Strong analytical judgment with the ability to manage multiple concurrent assessments, prioritize effectively, and drive risk‑based decisions…