As Senior Manager in Model Risk Management, responsible for implementing compensating controls and mitigating risks associated with machine learning models. Through validation, monitoring, documentation, and collaboration with key stakeholders, the work enhances model reliability and effectiveness while ensuring compliance with regulations and industry standards.
LLM Validation & Model Risk Governance
Developed and validated LLM solutions — BART, GPT, and Llama variants — for deployment in financial services contexts. Validation work covered performance benchmarking, distributional shift detection, and documentation aligned with SR 11-7 model risk governance requirements.
Led a BART-based document summarisation initiative that reduced manual analyst review effort by 90%, replacing labour-intensive review cycles with validated automated summaries that preserved regulatory traceability.
Identity Threat Detection & Response
Engineered Identity Threat Detection & Response (ITDR) models to detect anomalous identity-based activity across enterprise systems. Models were designed for low false-positive operation in high-stakes financial environments, with continuous monitoring pipelines and threshold tuning against adversarial baselines.
Adversarial Robustness & Security Testing
Executed comprehensive validation test suites for adversarial robustness and vulnerability assessments of AI systems — covering prompt injection, jailbreak resistance, data poisoning scenarios, and model extraction attacks. Findings directly informed model compensating controls and risk acceptance documentation presented to senior stakeholders.