Lead Data Analyst/Data Scientist – Differential Privacy

About the role

E-Logic, Inc. is seeking a Lead Data Analyst/Data Scientist specializing in Differential Privacy to support the IRS RAAS Statistics of Income (SOI) Division in developing, evaluating, implementing, and optimizing privacy-preserving methodologies for Qualified Opportunity Zone (QOZ) statistics. The position will lead the analytical and statistical aspects of the project, including differential privacy methodology, statistical disclosure limitation, privacy-utility analysis, synthetic data development, statistical modeling, evaluation of privacy parameters, protected statistical outputs, technical documentation, stakeholder support, training, and knowledge transfer.

Responsibilities

• Develop, test, implement, and evaluate statistically valid methodologies for producing aggregated QOZ statistics while protecting taxpayer information.
• Apply statistical disclosure limitation methodologies based on differentially private algorithms implemented in Python.
• Evaluate and apply differential privacy techniques for statistical disclosure control, privacy-preserving computation, and secure aggregation.
• Evaluate and pilot the Tumult Analytics Python framework or a comparable privacy-preserving analytics framework.
• Develop and refine privacy mechanisms and parameters while balancing privacy protection and statistical utility.
• Evaluate privacy budgets, including epsilon-level scenarios and the impact of reporting granularity on error and bias.
• Analyze IRS tax and administrative datasets associated with Opportunity Zone investments and Qualified Opportunity Funds.
• Support analysis involving relevant Form 8996 and Form 8997 records.
• Conduct statistical analyses of QOZ and QOF data.
• Support production of aggregated statistics for Government analysis and public reporting.
• Develop and assess methods for releasing datasets such as QOF counts, QOF investment amounts, and percentages of QOZ-designated census tracts while maintaining FTI confidentiality.
• Develop synthetic test data reflecting the structure and statistical characteristics of underlying population data.
• Develop an evaluation application capable of estimating error rates and bias metrics across different epsilon levels and reporting granularities.
• Quantify privacy-utility trade-offs and provide decision-support information for selecting appropriate privacy parameters.
• Validate statistical outputs and analytical results.
• Generate preliminary visualizations and summary tables demonstrating the feasibility of compliant QOZ statistical outputs.
• Develop a modular, differentially private analytical application that can be reused and adapted for future IRS and Treasury statistical reporting projects.
• Support implementation of the Year 1 statistical modeling framework using TY 2026 data in Year 2.
• Support production-level analytical pipelines and automated statistical output generation.
• Review and validate methodologies, models, privacy parameters, and analytical outputs.
• Refine statistical methodologies based on stakeholder feedback, data characteristics, and Treasury guidance updates.
• Produce transparent, reproducible analytical code and documentation consistent with Federal statistical standards.
• Contribute to data exploration reports, data dictionaries, technical assessments, methodology documentation, QA documentation, and technical reports.
• Document analytical methods, assumptions, privacy mechanisms, testing procedures, results, limitations, and privacy-utility trade-offs.
• Support preparation of documentation suitable for internal review, governance processes, OMB submission, and public release, as applicable.
• Provide structured training, workshops, code walkthroughs, and one-on-one mentoring for SOI analysts and data scientists.
• Develop user guides, technical documentation, troubleshooting procedures, instructional materials, and recorded training sessions.
• Support the transition of contractor-developed capabilities to fully self-sustaining Government operations.