Senior ML Engineer

Engagement 20-week non-production build on the customer's AWS landing zone. Start expected in the coming weeks; exact date to be confirmed. Allocation, rates, contracting terms and travel arrangements are to be agreed. Role purpose Deliver the machine-learning component of Aperam's Order Tolerance Management solution alongside AWS Professional Services. Build and validate tabular classification models that recommend ways to resolve order-tolerance problems, with clear explanations for business users. Responsibilities • Prepare, clean and validate model input data on Amazon S3. • Develop tabular classification models using XGBoost and/or LightGBM; implement validation, probability calibration and SHAP explainability. • Implement workflows using SageMaker Processing, Training, Pipelines, Model Registry, Batch Transform and Clarify. • Integrate Amazon Bedrock to produce plain-language explanations of recommendations. • Collaborate with the application engineer on prediction outputs and application integration. • Contribute reproducible code and documentation using the agreed infrastructure and CI/CD tooling. Keep MLOps proportionate to Phase 1, where it is explicitly light. Required experience • Proven personal ownership of tabular ML delivery on Amazon SageMaker, supported by project-level examples. • Strong Python skills and experience preparing data on S3. • Hands-on XGBoost/LightGBM modelling, model validation, calibration and explainability. • Practical experience with the SageMaker services listed above, including batch inference and model registration. • Ability to explain personal contribution, data/model scale and measured outcomes clearly. Technology environment AWS and Python; SageMaker; S3; Bedrock; Terraform; Azure DevOps CI/CD; Docker/ECR. The wider engagement also uses Entra ID. What to include with your application Include project examples naming the SageMaker services used, your own implementation responsibilities, dataset scale, validation approach, calibration and explainability work, and measured outcomes. General AWS, data-engineering or GenAI experience alone does not establish the core requirement. Expected outputs Recommended delivery outputs, subject to agreement: reproducible preparation/training and batch-inference workflows; registered model versions; evaluation and calibration evidence; explainability outputs; integration documentation and handover. Scope and status Based on the AWS delivery technical lead's written requirements of 23 October 2026. Seniority and the expected-output wording are Cloudworkz recruiting recommendations, subject to confirmation. This brief is not a contract or an offer of employment