Research Engineer, Sandboxes

Mistral is looking for a Research Software Engineer to join the Sandboxes team, onsite in London, ZH. The role centers on building and operating a platform that runs isolated, untrusted code at scale, across multiple Kubernetes clusters and data centers. This is a hands-on engineering position in a small, senior team where the emphasis is on clean, reliable systems code that scales, not on ML research. If you enjoy tackling infrastructure challenges at scale and shaping how researchers use a platform, this could be a great fit.

They provide full-stack AI solutions for enterprises across finance, manufacturing, defense, healthcare and the public sector, co-developing AI systems that clients can run on their own terms. The team works across Europe, North America, Asia and the Middle East, and they describe the environment as fast-moving and team-spirited, with a clear focus on practical impact and ownership. The company notes that the wrong hire is worse than no hire, so they prize thoughtful, capable engineers who can drive features from idea to production.

The Sandboxes role sits at the crossroads of distributed systems, low-level virtualization and platform design. You’ll be shaping how researchers interact with the sandbox, while maintaining performance and security guarantees. The setup in London, ZH is onsite, and you’ll be part of a small group that ships high-impact infrastructure used by data generation and RL training workflows.

Inside the sandbox backbone

You will design, build, and run a sandboxing platform that executes untrusted code at scale across several Kubernetes clusters, spanning both bare-metal data centers and public cloud environments. The effort combines engineering rigor with practical product thinking to deliver a reliable, scalable runtime.

The sandbox runtime will be developed with Firecracker to achieve quick startup times, efficient snapshots, and scheduling that keeps hardware usage lean. A key thread is reducing latency and keeping overhead low so that millions of sandboxes can operate concurrently without blocking essential resources.

Expect to care deeply about performance, reliability, and observability; code you write in Rust and Python should be robust and well-tested, with CI/CD and monitoring baked in. You’ll own cross-cluster networking, define isolation guarantees, and keep an eye on security boundaries and anomaly detection to spot suspicious activity.

Collaboration with researchers and ML engineers is routine, translating their needs into platform features that streamline synthetic data generation and RL training within the sandbox. The role blends hands-on infrastructure work with a user-facing sensitivity to how the platform gets used in real research settings.

Must-have ingredients for this role

Here's what the team needs from a candidate who can thrive in this role:

• A solid background in software engineering and distributed systems, with hands-on experience building and operating large-scale infrastructure.
• Deep expertise in at least one of the following areas: Kubernetes and cluster operations, cross-cluster networking, observability, or low-level virtualization.
• Fluency in Python or Rust, and comfort writing systems code where performance and security are important.
• Strong software-design instincts, including modularity, testing, CI/CD, observability, and a bias toward reliability.
• A self-starter mindset with fast learning ability; core skills matter more than specific tools.
• Low-ego and a practical, team-oriented approach that works well in a small, high-ownership environment.

Nice-to-haves

• Experience with Firecracker, microVMs, or similar virtualization tech.
• Product-mindedness, with an understanding of how sandboxes are used for synthetic data generation or RL training.
• Experience with GPU clusters, CUDA, or ML workloads.
• Prior SRE or platform engineering experience at scale.
• Open-source contributions or widely used internal tools.

Pay, benefits and practical details

Mistral offers a broad benefits package aimed at supporting well-being, growth, and work-life balance. The exact mix varies by country and can include healthcare coverage, parental leave, retirement plans, relocation support, wellness programs, meal and transportation allowances, and other location-specific perks. For the most up-to-date details in your location, you can view the Benefits page.

Your privacy matters. You can learn more about how personal data is handled in the Applicant Privacy Policy.

Advice for applicants

Lead your resume with your distributed-systems and infra experience, especially where you built or operated large-scale platforms and worked across Kubernetes clusters or virtualization layers.

Show concrete examples that prove you can meet the must-have skills. If you’ve developed scalable runtimes or worked on security boundaries and cross-cluster networking, call those outcomes out with metrics or concrete outcomes.

Be ready to discuss system-design ideas around sandbox isolation, Firecracker usage, and latency optimization. You may be asked how you would balance fast startup with strong security guarantees, or how you monitor for suspicious activity across a multi-cluster setup.

Consider asking a precise, product-focused question at the end, such as how researchers currently use sandboxes for synthetic data generation or RL training, and what platform features are planned next to support those workflows.