Research Engineer — Privacy & Security Differential privacy, leakage tests, and PETs that survive a production train. Remote / Hybrid · →

Meramia Technologies; Dubai; Remote; AED 300,000 - 540,000; Full time

Job summary

Meramia Technologies in Dubai seeks a researcher-strong ML privacy engineer to prototype and land privacy-enhancing algorithms such as differential privacy, secure aggregation, and federated learning on our training and inference stacks.

You will red-team models for membership inference, inversion, and data memorisation, build evaluation suites for ML engineers, and work with research, platform, security, legal and product to embed privacy as a guardrail in code while exploring the

Qualifications

• Proficiency in privacy-preserving ML techniques
• Ability to prototype privacy algorithms in Python
• Experience turning papers into tested modules

Responsibilities

• Prototype privacy-enhancing algorithms on training and inference stacks
• Red-team models for membership inference, inversion, memorisation
• Build evaluation suites for ML engineers across a model life
• Collaborate with research, platform, security, legal and product to embed privacy as guardrails
• Investigate privacy–utility trade-offs: capability, latency and cost

Skills

Python

Differential privacy

Federated learning

DP-SGD

Model red-teaming

Privacy explainability

Tools

PyTorch

JAX

Job description

The seat sits between the models we train and serve and the data those models must not memorise — deep learning research, systems work, and the privacy techniques that have to hold when the estate is a client's.

• Prototype and land privacy-enhancing algorithms — differential privacy, secure aggregation, federated learning — on the training and inference stacks we actually run.


• Red-team the models for membership inference, inversion, and training-data memorisation. Write the failure down before someone else finds it.


• Build the evaluation suites and diagnostic libraries ML engineers can use across a model's life, not a one-off notebook.


• Sit with research, platform, security, legal and product so a regulation or a security principle becomes a guardrail in code.


• Investigate the privacy–utility trade: capability, latency and cost against a guarantee you can state.

What you will do

• Prototype and land privacy-enhancing algorithms — differential privacy, secure aggregation, federated learning — on the training and inference stacks we actually run.


• Red-team the models for membership inference, inversion, and training-data memorisation. Write the failure down before someone else finds it.


• Build the evaluation suites and diagnostic libraries ML engineers can use across a model's life, not a one-off notebook.


• Sit with research, platform, security, legal and product so a regulation or a security principle becomes a guardrail in code.


• Investigate the privacy–utility trade: capability, latency and cost against a guarantee you can state.

What you bring

• PyTorch or JAX, and research-grade Python you will test rather than demo.


• Differential privacy (including DP-SGD), secure multiparty computation, or federated learning — implemented, not only cited.


• The attack surface: extraction, membership inference, poisoning — and how you measured it.


• A paper you can turn into a well-tested module without losing the claim that made the paper worth reading.


• The ability to explain a mathematical privacy guarantee to an engineer and to a policy lead in the same week.

Useful, not required

• Peer-reviewed work or open-source in privacy, security, cryptography or machine learning — NeurIPS, ICLR, USENIX Security, IEEE S&P, or the equivalent venue.


• PETs on a distributed train or a high-throughput inference path, not only on a single node.

The first quarter

• Audit the privacy evaluations and training workflows already in use.


• Ship one automated evaluation or differential-privacy module into the internal ML path.


• Write the privacy–utility result for the architectures we serve, with a deployable recommendation.