Anh Phong Tran

This role focused on validating AI/ML tools used across all areas of the bank with a focus on risk management, proper use of the algorithms, and educating stakeholders as well as peers on the newest developments in the area of AI.

• Evaluated AI/ML models in the areas of NLP, Computer Vision, and Fraud Detection. Main projects involved tagging audio conversations with LLMs, clustering complaints and feedbacks with word/sentence embeddings, and identifying suspicious trades using anomaly detection techniques.
• In Credit Loss Forecasting, pioneered a novel internal KPI metric to assess performance breaches in weighted portfolios. Also uncovered a critical issue leading to substantial overestimation in EAD forecasting for a $360 billion portfolio.
• Delivered department-wide educational content on machine learning applications, covering anomaly detection in unsupervised learning, convolution neural networks for computer vision, and NLP related topics
• Developing a use case of applying GenAI, GPT-4 in the validation workflow for Model Risk Management

In this role, I assisted and led various research cutting-edge projects in the area of Medical Physics using a diverse set of mathematical models and techniques to uncover ways to improve cancer treatments.

• Conducted pattern recognition of oscillatory time-series data showing the impact of p21 and p53 dynamics on cell cycle arrest. This research culminated in a published article in Nature’s Cell Death \& Differentiation journal.
• Invented innovative methods for patient classification using multi-omics data resulting in 2+ publications using Wasserstein curvature on gene networks and preconditioned random forest regression for genomics data analysis.
• Sped up 100X an algorithm to calculate optimal mass transport of fluid flow in glymphatic systems by implementing customized multi-core parallelization of the code and code optimization.