Hybrid Algorithms Engineer, Differential Privacy

at Oblivious

• Job title: Algorithms Engineer, Differential Privacy
• Job type: Full Time
• Experience level: Mid level, Senior
• Salary: Not specified
• Degree requirement: No Education Requirement
• Tech skills: Python, SQL, Tensorflow
• Location requirements: Hybrid, Dublin, Ireland

About the role

• Algorithms Engineer designing core components of differential privacy systems for Oblivious. Translating mathematical theory into production-ready code and solving privacy challenges.

Responsibilities

• Privacy Accounting & Mechanisms: Implement and analyse privacy loss accountants (RDP, zCDP) and their conversions to (ϵ, δ)-DP. Calibrate and apply noise mechanisms (Gaussian, Laplace) based on rigorous sensitivity analysis.
• Differentially Private SQL Engine: Develop algorithms for static and dynamic sensitivity analysis of relational operators. Build query rewriting logic to inject calibrated noise and manage a per-user privacy budget ledger.
• Compiler & Static Analysis: Use Python AST manipulation and static analysis to enforce a DP-safe execution environment, ensuring user-submitted code cannot leak private information.
• DP Synthetic Data: Implement and benchmark state-of-the-art algorithms (e.g., MWEM, PGM, PrivBayes variants) for high-dimensional synthetic data generation, analysing their privacy-utility trade-offs.

Requirements

• Strong foundation in probability, statistics, and linear algebra. You must be comfortable with statistical modelling, proving bounds, and reasoning about error/variance.
• Proficiency in Python for scientific computing, including numerical stability considerations (e.g., floating-point precision, clipping, scaling).
• Demonstrated ability to translate mathematical concepts from academic papers or technical specifications into robust, well-tested code.
• **Desirable**
• Direct experience with differential privacy concepts or libraries (OpenDP, SmartNoise, TensorFlow Privacy).
• Knowledge of compiler design, abstract syntax trees (ASTs), or program analysis.
• Experience with machine learning, particularly with noise models, statistical learning theory, or generative models.
• Familiarity with SQL parsers or database internals.

Benefits

• Private health insurance and pension plan
• Paid time off
• Training & development