AI Infrastructure Engineer: High-Scale Sandboxing & RL

Socket.dev; Palo Alto (CA); On-site; USD 210,000 - 260,000; Full time

Benefits offered by this job

Healthcare coverage

Relocation support

Wellness programs

Job summary

Mistral is seeking a seasoned Systems/Distributed Infrastructure Engineer to lead end-to-end execution and data infrastructure powering its agentic models and coding assistants. You will help design scalable synthetic data generation, ultra-fast training environments, and high-throughput execution engines across hybrid and multi-cloud clusters.

You will tackle challenges from 1 million concurrent sandboxes to optimize training codebases, pipelines, and orchestration, working with researchers to

Qualifications

• 4+ years in Systems Engineering/Distributed Systems/Cloud Infra or MLOps.
• Experience building high-throughput data processing pipelines and generation pipelines for large-scale datasets.
• Strong proficiency in Python, Go, C++ or Rust; experience profiling and optimizing high-performance ML/backend code.

Responsibilities

• Design, deploy and operate high-throughput sandboxing for untrusted code across 1M+ environments.
• Architect data generation pipelines for synthetic code, agent trajectories, rollouts and self-play data.
• Optimize agent training codebases and distributed runtimes (Python, Go, C++, Rust) to improve GPU utilization and reduce rollout overhead.
• Reduce sandbox cold-start times with container warm pools and efficient image delivery across hybrid clusters.
• Develop Kubernetes-native controllers and queuing systems for diverse hardware fleets.
• Ensure multi-tenant isolation and secure boundary enforcement across runtimes and networks.

Skills

Distributed Systems

Cloud Infra

MLOps

Kubernetes

Container Tech

Python

Go/C++/Rust

4+ years experience

Tools

Kubernetes

Docker

CRIU

Firecracker

gVisor

Job description

Mistral is seeking a seasoned Systems/Distributed Infrastructure Engineer to lead end-to-end execution and data infrastructure powering its agentic models and coding assistants. You will help design scalable synthetic data generation, ultra-fast training environments, and high-throughput execution engines across hybrid and multi-cloud clusters.

You will tackle challenges from 1 million concurrent sandboxes to optimize training codebases, pipelines, and orchestration, working with researchers to