Tomas Romo

Led end-to-end AI/ML initiatives for clients focusing on fraud detection, risk modeling, and customer personalization, delivering scalable and compliant solutions. Architected real-time ML pipelines utilizing AWS services including S3, SageMaker, Lambda, and Kinesis, ensuring high availability and low-latency processing.

Developed state-of-the-art NLP models to automate dispute classification and extract customer interaction insights, improving resolution times and accuracy. Implemented deep learning architectures with PyTorch, including attention-based models, to enhance fraud prediction capabilities on complex transaction data.

Led the architecture and development of real-time fraud detection systems using gradient boosting frameworks such as XGBoost and LightGBM, achieving high precision and recall in live environments. Designed and deployed scalable ML infrastructure leveraging GCP, Kubernetes, Apache Airflow, and Docker, supporting rapid iteration and continuous deployment.

Built and validated advanced machine learning models for fraud detection and risk assessment, leveraging Python, scikit-learn, and Spark to improve accuracy and scalability. Designed and implemented predictive scoring systems for merchant credit risk and transaction anomaly detection, increasing early risk identification rates.

Studied Computer Science at the University of Texas - Austin. Gained foundational knowledge and practical skills in various aspects of computer science.