Andrew Saeed

Architected and deployed real-time fraud detection models using ensemble techniques. Designed and managed MLOps pipelines with Kubeflow, MLflow, Jenkins, ensuring continuous training and delivery. Led the migration of 20+ legacy models to AWS SageMaker. Developed NLP systems using BERT/RoBERTa for secure document classification in finance and legal domains. Built model drift monitoring with CloudWatch, custom alerts, and retraining triggers integrated with Airflow DAGs. Conducted model governance audits and ensured compliance with AI ethics, model transparency, and GDPR standards. Mentored and led a cross-functional team of 7, including ML Engineers, Data Scientists, and DevOps engineers. Presented strategic AI initiatives to executives, aligning technical delivery with OKRs and revenue-impact KPIs. Led integration of SHAP and LIME for explainability and co-authored internal AI Ethics framework focused on fairness, reproducibility, and bias detection.

Built scalable ML pipelines for customer segmentation, predictive scoring, and recommendation engines using collaborative filtering. Engineered end-to-end ETL pipelines on GCP using Apache Beam and BigQuery, enabling real-time analytics. Designed deep learning computer vision models (YOLOv5, EfficientNet) to automate manufacturing defect detection. Developed feature stores and version control systems using DVC, integrated with GCP Data Catalog. Reduced model drift with online learning, real-time monitoring, and concept drift detection algorithms.

Developed object detection pipelines using YOLOv3 and EfficientNet for industrial visual inspection tasks. Built recommender systems with matrix factorization and gradient boosting. Automated ETL and streaming data pipelines using BigQuery and Apache Beam; integrated with DataFlow and Pub/Sub. Led ethical AI development for customer-facing ML tools; implemented model explainability dashboards. Managed deployment of production models using Azure ML endpoints and version control with Git and CI/CD pipelines. Conducted A/B testing for personalization models, enabling performance monitoring across key business segments. Presented use case outcomes to UK-based enterprise clients in retail and manufacturing. Collaborated with product managers to scope ML deliverables within agile sprints.