Hiring nowPrivacy-preserving and confidential AI

Privacy-Preserving ML Researcher

Also called Private ML Scientist

A Privacy-Preserving ML Researcher builds learning methods that limit what a model can reveal about the data behind it. That lets a company train and serve on sensitive data with stated, testable limits on leakage.

What they do

The work starts with a threat model: who the adversary is, what they can see, and what counts as a leak. From there the researcher designs or adapts a learning method, often built on differential privacy, to meet that limit. They run privacy attack suites against their own models to measure real leakage. They test how much accuracy is lost and look for ways to recover it. They work with security and platform engineers on confidential compute and cryptographic components. They write up assumptions clearly so others can check them.

What a strong hire delivers

In the first months, a strong hire produces a written threat model the team agrees on. They deliver a working private training method, with a utility evaluation set against a non-private baseline. They add empirical leakage tests that run repeatedly, not once. They also leave the company with a clear view of which privacy claims it can make to customers and which it cannot.

How you can tell they are good

Ask them to walk through a past method from threat model to result. A strong candidate states the assumptions plainly, explains where the guarantee stops, and shows the utility cost without being asked. Ask how they chose their privacy parameters and how they tested for leakage. Be wary of anyone who treats a guarantee as a blanket promise or reports only the best result.

Why it is hard to fill

Few people combine formal privacy theory with hands-on model training. Most come from one side. Cryptographers, security engineers and ML researchers all use different titles and vocabularies, so job searches miss them. Many have academic or research-lab careers and are not looking. Candidates who can also explain trade-offs to a non-specialist founder are rarer still. Expect a thin, slow-moving market.

Where to find them

Look in privacy engineering, applied cryptography, security research and ML research groups. Titles include Private ML Scientist, Privacy Engineer, Research Scientist and Security Researcher. They come from frontier labs, large platform companies, health and finance technology firms, confidential compute providers and university research groups. Open-source privacy library contributors and authors of attack papers are also worth a look.

Indicative pay

Not enough published evidence yet to give a fair range for this role, which is common for work this new. We benchmark it for you as part of a search.

Interview questions for a Privacy-Preserving ML Researcher

  1. Describe a threat model you wrote for a learning system. What did you include, and what did you deliberately leave out?

    A strong answer names the adversary, their access and the leak being prevented, then states what is out of scope and why. Worry if the answer is vague, or if the threat model was written after the method.

  2. How have you measured the utility cost of a private method, and how did you decide the cost was acceptable?

    Look for comparison with a non-private baseline, clear metrics and a link to the product need. Strong candidates discuss tuning and trade-offs. Worry about cherry-picked results or no baseline at all.

4 more questions, with what a strong answer covers, are in the Role Kit.

Job description you can post

We build AI products that handle sensitive information, and we need a researcher who can limit what our models reveal. You will define the privacy risks, develop learning methods that address them, and show with evidence how well they work.

What you will do

  • Write explicit threat models for training and inference systems
  • Develop and adapt private learning methods, including differentially private training
  • Evaluate the trade-off between privacy and model utility against clear baselines
  • Run privacy attacks on our models and track empirical leakage
  • Work with security and platform engineers on confidential compute and cryptographic components
  • Document assumptions and limits so privacy claims can be reviewed

What you will bring

  • Demonstrated work building or evaluating differential privacy or other private learning methods
  • Experience writing threat models and stating guarantees with their assumptions
  • Hands-on model training and evaluation experience
  • Practical use of privacy attack techniques to test leakage
  • Background in privacy engineering, security, cryptography or ML
  • Clear written communication of technical limits to non-specialists

Nice to have

  • Experience with confidential compute or cryptographic libraries
  • Published or open-source work on privacy or attacks
  • Experience with formal verification or data provenance

Free to use and adapt. If it saves you time, a link back to this page is appreciated.

Skills that matter

  • Privacy engineeringAdvancedLimit disclosure and unauthorised personal-data use
  • Security engineeringAdvancedProtect systems, identities and sensitive assets
  • Formal verificationWorkingProve stated properties under explicit assumptions
  • Data quality and provenanceWorkingTrack dataset origin, validity and transformations

Other roles in privacy-preserving and confidential AI

EmergingConfidential AI EngineerA Confidential AI Engineer runs model training and inference inside attested, restricted execution environments. They let a company use sensitive…EmergingAI Privacy Attack ResearcherAn AI Privacy Attack Researcher tests, with authorisation, whether a model or its data can leak sensitive information. They show the company where…Hiring nowAI Privacy EngineerAn AI Privacy Engineer builds consent, minimisation and access controls into AI workflows, so sensitive data is not exposed in training or…

Related roles in other families

EmergingAI Control ResearcherAI control, monitoring and containmentAn AI Control Researcher designs the safeguards that limit what a model can do if it turns out to be unreliable or hostile. They keep risk bounded…EmergingAgent Containment EngineerAI control, monitoring and containmentAn Agent Containment Engineer builds the boundaries that keep AI agents from doing harm, even if the model misbehaves or is attacked. They matter…EmergingAI Monitor Evaluation EngineerAI control, monitoring and containmentAn AI Monitor Evaluation Engineer tests whether the systems that watch model behaviour actually catch dangerous or unauthorised actions. It matters…EmergingAI Shutdown and Recovery EngineerAI control, monitoring and containmentAn AI Shutdown and Recovery Engineer builds and tests the ways to stop an AI system mid-action and bring it back to a known good state. It matters…

How we wrote this

Role definitions are our own editorial work, checked against occupational classifications, standards, research and employer pages. Titles vary by company, and this is a description of the work, not a job advert.

← Privacy-preserving and confidential AIAll new job families

Start a search

Hiring a Privacy-Preserving ML Researcher?

Tell us the work. We find the people already doing it, whatever their job title says, and you meet them already interviewed for your role.