Note: The job is a remote job and is open to candidates in USA. Dice is a recruiting company offering a fully remote contract opportunity in a federal environment. The Data Scientist III will develop and validate predictive, prescriptive, anomaly detection, and optimization models on a greenfield data and AI platform, while ensuring model integrity, explainability, privacy, fairness, and compliance with government security standards.

Responsibilities

• Build and validate risk-scoring models using synthetic tabular data and engineered features from curated datasets
• Develop anomaly and outlier detection techniques to identify irregularities in records and processing data streams
• Design and implement optimization models for prioritization, routing, and resource allocation tasks
• Ensure rigorous model validation, including calibration, discrimination, stability, and explainability assessments
• Package and document model deliverables as jobs and Asset Bundles, tracked in MLflow, including assumptions, limitations, and revalidation needs
• Support privacy-preserving synthetic data generation, maintaining distributional fidelity and re-identification risk management
• Collaborate with multi-disciplinary teams to incorporate model feedback and improve model performance
• Prepare comprehensive documentation, including assumptions and limitations, to facilitate knowledge transfer and audit readiness
• Conduct fairness and adverse impact analysis within regulated decision-support environments
• Maintain compliance with government security standards such as FedRAMP, NIST 800-171, and CUI handling requirements

Skills

• This position requires eligibility for a U.S. Government security clearance. In accordance with federal law, U.S. citizenship is required (active T5/SSBI federally adjudicated security clearance
• Hands-on experience with Databricks, Python, SQL, and Spark in a production environment
• Proven expertise in feature engineering on tabular and time-series data, including encoding, aggregation, and leakage prevention
• Strong background in supervised learning methods such as gradient boosting (XGBoost, LightGBM) and regularized regression
• Knowledge of model calibration, evaluation under class imbalance, and explainability tools like SHAP
• Experience with anomaly detection techniques such as isolation forests, autoencoders, or statistical process control, including validation without labels
• Proficiency in optimization techniques using LP/MIP or heuristic methods like OR-Tools, Pyomo, or SciPy
• Experience generating privacy-preserving synthetic data from CUI, PII, or similar source data, ensuring data utility and security
• Strong written and verbal communication skills, with the ability to document complex models and collaborate across technical and non-technical teams
• Self-motivated with the ability to execute independently against fixed milestones with minimal oversight
• Government or defense contracting experience is preferred
• Familiarity with FedRAMP, NIST 800-171, CMMC L2, or CUI handling is advantageous

Benefits

• This is a fully remote role and can be performed from any approved location within the United States.

Company Overview