About This Role
Working with teams to transform design data and historical builds into actionable planning outputs and product part suggestions is the core of this Machine Learning Engineer role. You will also manage the evaluation loop to ensure these outputs are reliable enough for approval. The modeling work extends beyond large language models, focusing on complex domains with sparse data, making it ideal for an engineer who is systematic in error analysis and eager to deeply understand technical domains from experts.
Responsibilities
- Partner directly with planners and manufacturing engineers to understand current build artifact production and define correct output before building
- Design and ship models and pipelines that produce planning outputs from specifications and comparable prior builds
- Build evaluation harnesses, error taxonomies, and regression checks to measure output quality against expert judgment, and drive accuracy up through systematic error analysis
- Structure and normalize engineering and production data across PLM, ERP, and design systems, including part resolution and comparability across prior builds
- Deliver outputs into human review and approval workflows with traceability, so that they can be verified and corrected before use
- Stand up models and services in a restricted enterprise environment, with lightweight CI/CD, monitoring, and repeatable deployment
Requirements
- 5+ years of software engineering experience, including at least 3 years building and shipping applied machine learning systems in production
- Strong Python, along with C++ and SQL, and regular hands-on use of Pandas, Scikit-learn, XGBoost, and LightGBM
- Proficiency with at least one deep learning framework (PyTorch, TensorFlow, or Keras), including models you have trained and tuned yourself, paired with a strong understanding of ensemble methods such as bagging, boosting, stacking, and blending
- Demonstrated experience applying ML in a low-data, expert-labeled domain such as defense, aerospace, medical, legal, or industrial engineering, where correctness is defined by subject matter experts rather than by an abundant labeled dataset
- Rigorous evaluation practice, with a track record of building evaluation harnesses and using structured error analysis to improve system accuracy over successive iterations
- Experience working with complex structured and semi-structured enterprise data, including entity resolution, hierarchical or graph-shaped records, and messy master data
- Sound judgment about data handling and deployment constraints in restricted environments, including what can and cannot cross a system boundary
- Outstanding communication skills, a bias for action, and the ability to deliver results independently in ambiguous environments with direct stakeholder contact