Latest AI JobsBrowse Jobs
Mistral AI logo

Research Engineer, Code Agents Infra

Mistral AIPalo AltoScience
Apply on company site
Salary
Not listed
Job type
Full-time
Work mode
Remote
Posted
4 Aug 2026
Status
Open · checked 7d ago

About this role

Role Overview

The Research Engineer, Code Agents Infra position at Mistral AI is a full-time science department role based in Palo Alto with remote options available. This position focuses on designing and operating the comprehensive execution, training, and data infrastructure that drives the company's agentic models and coding assistants.

Key Responsibilities

Build, operate, and scale high-throughput sandboxing platforms that safely execute over 1 million concurrent instances of untrusted code generated by large language models for evaluation and reinforcement learning.
Architect robust data pipelines capable of managing synthetic code generation, agent trajectories, rollouts, and self-play data collection to support post-training loops.
Optimize agent training codebases and distributed execution runtimes using technologies like PyTorch, Ray, SLURM, and Kubernetes to reduce rollout overhead and maximize hardware efficiency.
Lower sandbox cold-start durations down to sub-second levels via container warm pools, snapshot tools, and tailored image delivery layers across hybrid clusters.
Develop Kubernetes-native custom controllers, CRDs, and queuing systems to intelligently route transient evaluation, synthetic data, and agent execution workloads across diverse hardware fleets.
Enforce stringent network and process isolation for untrusted agent tasks utilizing specialized container and sandboxing runtimes such as gVisor and Firecracker.
Maintain high availability, telemetry, and automated self-healing across millions of short-lived tasks while taking part in on-call rotations for crucial infrastructure.

Requirements

Minimum of 4 years of professional experience working in Systems Engineering, Distributed Systems, Cloud Infrastructure, or MLOps tailored to LLM and reinforcement learning workloads.
Demonstrated background in data and pipeline engineering, constructing high-throughput data processing systems for large datasets using Ray, Spark, or custom distributed queues.
Extensive familiarity with Kubernetes and container technology, including writing custom controllers, managing Linux cgroups and namespaces, and optimizing container image layers.
Advanced software engineering capabilities in Python, Go, C++, or Rust, accompanied by a history of profiling and refining high-performance machine learning or backend systems.
Practical, hands-on exposure to lightweight virtualization, container runtimes, or WebAssembly like Docker, gVisor, or Firecracker.
Deep understanding of task queue systems, resource schedulers, and low-latency queuing setups designed for high-volume, transient tasks.
High comfort level with ambiguity and a strong desire to collaborate closely with AI researchers to translate experimental ideas into production systems.

Skills and Tools

Python
PyTorch
Kubernetes
Docker
Rust
Spark
C++
LLM
MLOps

About the Company

Mistral AI delivers full-stack artificial intelligence solutions ranging from frontier models and developer tools to custom applications and compute power. The company collaborates with enterprises across critical fields like finance, healthcare, manufacturing, defense, and the public sector to deploy customized AI systems. Operating as a distributed, collaborative team across North America, Europe, Asia, and the Middle East, Mistral AI fosters a low-ego, creative culture committed to driving technological innovation.

How to Apply

Interested candidates should submit their application directly through the official Mistral AI careers page.

Skills and tech stack

PythonPyTorchKubernetesDockerRustSparkC++LLMMLOps

Questions candidates ask

What is the location of the Research Engineer role?

The role is based in Palo Alto with remote flexibility.

How many years of experience are required?

Candidates must have 4 or more years of experience in Systems Engineering, Distributed Systems, Cloud Infrastructure, or MLOps supporting LLM and RL workloads.

What department does this role belong to?

This position is part of the Science department at Mistral AI.

What core technologies and programming languages are expected?

Applicants should be proficient in languages such as Python, Go, C++, or Rust, and have experience with PyTorch, Kubernetes, Docker, Spark, Ray, and sandboxing tools like gVisor or Firecracker.

Where should I submit my application?

Applications should be submitted directly on the official Mistral AI careers website.

Ready to apply?

The application is on the official Mistral AI careers page.

Apply on company site