This project will create an end-to-end, AI-powered email delivery system for a U.S. client. All work must be carried out on-site or remotely from New York State; residency is a strict requirement because of data-handling policies. Scope of work The core task is to design, train, and deploy a regression-based machine-learning model that predicts optimal email-send parameters from rich transactional data. I will supply secure access to the historical transaction logs; you will architect the data pipeline, engineer the relevant features, and iterate on model performance until it is production-ready. Once validated, the model should be wrapped in an API and integrated into our existing email infrastructure so it can trigger personalized sends automatically. Key responsibilities • Data ingestion and cleaning for large transactional datasets • Feature engineering geared toward regression targets (e.g., best send-time, expected revenue) • Model training, hyper-parameter tuning, and A/B test design • Deployment to a scalable environment (AWS or GCP preferred, but stack is flexible) • Documentation and handoff, including monitoring dashboards and retraining scripts Acceptance criteria A working API must return real-time predictions within agreed latency limits, integrate seamlessly with the current SMTP/ESP workflow, and include logging for compliance review. Final delivery is considered complete when the system runs in production and all documentation passes peer review. Tools & stack Python, scikit-learn or TensorFlow, SQL/NoSQL for data storage, and standard MLOps utilities (Docker, CI/CD) are anticipated, yet alternative libraries are welcome if they meet the same reliability and security standards. Timeline and milestones will be outlined together at project start, with code reviews scheduled at each major checkpoint.