Data Science Expert Oluwatosin Lawal unveils Fraud and Anti-Money Laundering detection strategies with Advanced Analytics

By Tosin Clegg

Oluwatosin Lawal, a business analyst and data science expert, is at the forefront of the transformation of anti-money laundering risk management through the integration of predictive models and large language models. His work tackles a persistent challenge in the financial industry: how to detect suspicious activity with accuracy and efficiency while reducing the noise of false positives that overwhelm compliance and investigation teams. By leveraging advanced machine learning models and large language models (LLMs), Oluwatosin is pushing the boundaries of how institutions identify, analyze, and mitigate risks in an era of rapidly evolving financial crime.

At his current role in a leading North American bank, Lawal operates within the anti-money laundering technology and analytics division where he is leveraging advanced analytics to help identify systemic gaps in fraud and anti-money laundering monitoring systems. His analysis goes beyond post-event evaluation by feeding insights back into the design and training of predictive models that complement existing rule-based systems. By coupling these insights with machine learning models, he is building a framework that significantly improves the detection of high-risk activity across the complex United States financial system.

A cornerstone of Lawal’s current work is the integration of LLMs into the transaction monitoring ecosystem. By applying natural language processing techniques to narrative fields, case notes, and unstructured regulatory data, he has demonstrated how LLMs can extract context that rules-based models and conventional ML algorithms often fail to capture. This allows for deeper risk understanding, such as linking customer intent to transaction flow, detecting subtle red flags in customer communications, and identifying recurring suspicious typologies across different jurisdictions.
Machine learning underpins much of Lawal’s technical approach. His projects involve deploying algorithms such as logistic regression, random forests, gradient boosting, and neural networks to evaluate transaction monitoring outcomes. Recognizing that AML data is highly imbalanced because fraudulent or suspicious activity represents only a fraction of the dataset, he applies resampling techniques and hyperparameter optimization to increase detection accuracy. This ensures that models are not only sensitive enough to flag true risks but also efficient in reducing false positives, thereby easing the operational burden on investigative teams and allowing them to focus on risk that prevents the American people from falling victims to fraud.

“One real challenge that a lot of top financial institutions face in financial crime detection is balance,” Lawal explains. “A system that misses suspicious activity exposes the institution to risk, but a system that flags too much irrelevant activity wastes valuable investigative resources. Machine learning and large language models give us the tools to strike that balance at scale, helping investigators streamline reviews of alerts and making it possible for the United States fraud victims to partially or fully recover lost funds”.

His ongoing work in integrating structured transactional data with unstructured intelligence sources represents a major advancement in AML detection. For instance, he has developed programs to validate and cleanse datasets feeding into monitoring systems, ensuring data integrity before modeling. He is also experimenting with fine-tuned LLMs capable of summarizing SAR narratives, clustering cases by typology, and generating investigative leads that streamline human decision-making. These innovations allow institutions to stay proactive in detecting and mitigating emerging threats and filing timely regulatory reports to the U.S. Department of the Treasury’s Financial Crimes Enforcement Network (FinCEN),
Lawal stated “The next frontier of fraud and AML risk detection lies in combining predictive modeling with natural language intelligence. LLMs have massive use case potential in financial institutions. Financial crime is dynamic, and to mitigate risks effectively, institutions must adopt models that not only analyze numbers but also interpret patterns in unstructured data such as case files and SAR narratives.”

Lawal’s career trajectory has prepared him uniquely for this space. With a bachelor’s degree in systems engineering, a master’s degree in statistical analytics, computing, and modeling, combined with multiple years working as a Data Scientist and Risk Consultant at a top consulting firm like KPMG, he built deep expertise in regulatory compliance reviews, developing systems for fraud detection and mitigation in financial services companies. This foundation of regulatory knowledge strengthens his ability to design models that align with compliance requirements while remaining innovative and practical.

Currently, he is also advancing efforts to consolidate data from legacy and new platforms into a centralized repository, creating a unified foundation for machine learning experimentation and deployment. This centralized hub not only enhances data accessibility but also provides the infrastructure for testing advanced LLM pipelines that can process billions of data points in near real time, which are emphasized in supervisory expectations under Bank Secrecy Act (BSA) and Office of the Comptroller of Currency (OCC) guidelines. His work in this area is positioning his institution to leverage AI-driven insights for both operational risk reduction and strategic compliance planning which helps to strengthen the integrity of the United States financial system.
The trajectory of Lawal’s work demonstrates more than technical innovation; it underscores a commitment to public good. His work not only address financial institutions’ operational efficiency but also advance broader regulatory objectives under the Bank Secrecy Act, FinCEN advisories, and U.S. national security priorities.

When asked for his long-term plans, Lawal stated “My vision for the future is to build advanced fraud and AML detection technologies that directly safeguard the financial security of the American people. Every instance of fraud or money laundering has ripple effects, draining family savings, threatening small businesses, and increasing costs across the economy. By leveraging machine learning and large language models to detect and prevent illicit activity, I aim to ensure that communities across the United States are better protected, trust in the United States financial system is strengthened. Every dollar kept out of the hands of criminals is a dollar preserved for the growth of the U.S. economy, community development, and the financial security of the American people.”

Through this blend of technical innovation, regulatory alignment, and human impact, Lawal is setting a precedent for how machine learning models and LLMs can be leveraged not just to meet compliance mandates for financial institutions, but to deliver tangible benefits to the American people.