Contract: 18-month fixed-term contract from signing (running to at least January 2028), with a view to extending or moving to permanent as a second round of programme funding is confirmed 

Location: Oxford, hybrid preferred, though we'd consider fully remote for the right person.

Oxfida (formerly Aioi R&D Lab – Oxford) is an AI R&D company based in Oxford, on a mission to harness AI to understand, predict, and manage risk, helping build a safer, more resilient society.

We sit at the intersection of academia and industry, working with Oxford's professors and researchers alongside commercial partners, and combining that expertise with real-world risk data at global scale to turn frontier research into AI that actually ships rather than just gets published.

Our work spans AI for insurance underwriting and claims, AI risk governance, privacy-preserving technologies for agentic AI, and quantum optimisation, alongside partnership projects on nature and supply chains and the emerging challenges nobody's solved yet.

We build AI products and solutions for insurers, businesses, and public-sector organisations worldwide, and run innovative research projects that push these technologies further, helping people make better decisions in an uncertain world.

Trusted AI. Built in Oxford.

You'd be joining CODAS, our flagship sovereign AI programme, built in partnership with Japanese public and private sector organisations and academic collaborators at Oxford and UCLA. The goal is a privacy-preserving generative and agentic AI ecosystem that protects personal and sensitive data by design, through data residency, anonymisation, secure deployment, and privacy techniques that hold up at both training and inference time.

As a Senior Machine Learning Scientist, you'll take technical ownership of research workstreams within CODAS. Rather than just executing a brief, you'll set the experimental direction yourself and push approaches like differential privacy, federated learning, and secure inference beyond what's already out there.

Responsibilities

• Own the technical direction of assigned CODAS research tracks, from framing the problem to validating the results.
• Design and build ML approaches for privacy-preserving generative AI, including differential privacy, federated learning, secure inference, and related techniques.
• Turn research into proof-of-concept systems and deployable components that feed directly into programme deliverables.
• Publish, contributing to papers, technical reports, and conference talks that put Oxfida's work in front of the wider AI community.
• Mentor other ML scientists and research associates working alongside you on the programme.
• Work directly with partner organisations to understand real technical constraints, not just theoretical ones.

Knowledge, Experience and Qualifications

• A Master's or PhD in Computer Science, ML, Statistics, Maths, or equivalent experience, plus 5+ years in applied ML/AI research or development.
• A genuine track record in privacy-preserving ML, generative AI, LLMs, or federated learning, since this isn't a "learn on the job" role.
• Strong Python engineering skills: ML frameworks, version control, CI/CD, reproducible experiments.
• Comfort working across the full ML lifecycle, from first prototype to something that can actually be deployed.
• Experience mentoring others and the communication skills to explain complex trade-offs to non-technical stakeholders.

Bonus points: differential privacy or secure computation experience, LLM/agentic AI/frontier AI evaluation work, published research, or experience in regulated industries (insurance, financial services, public sector).

Why join us

• Real technical ownership on a genuinely novel, high-profile programme, not a support role on someone else's research.
• Direct exposure to Oxford's academic network and CODAS's international partners, not just your immediate team.
• Work that ships. This isn't research for its own sake; it becomes part of a live sovereign AI ecosystem built with Japanese public and private-sector partners.
• This is a fixed-term contract tied to CODAS's current funding cycle, but there's a genuine intent to extend and move roles to permanent as a second funding round is confirmed. It isn't a contract designed to end.