Skip to content
All roles
Engineering · Sandbox & Runtime

Agent Infra Engineer (Sandbox & Runtime)

We’re building agent infrastructure for large model training, evaluation, and applications, enabling agents to complete complex tasks in code repositories, browsers, desktop software, and mobile apps — at scale and in reproducible environments. You’ll help design and build sandboxes, runtimes, and execution clusters, addressing reliability, resource isolation, task scheduling, and cost across diverse environments while exploring new technical approaches and platform capabilities as the business grows.

What you’ll do

  1. 01Design and build agent sandbox and runtime infrastructure, supporting environment creation, interactive execution, state management, persistence, recovery, and resource reclamation.
  2. 02Build scalable execution and resource scheduling capabilities. Improve concurrency management, resource quotas, prewarming, autoscaling, and failure recovery to continually improve system stability, task throughput, and cost efficiency.
  3. 03Evaluate and integrate suitable execution environments for code, browser, desktop, and mobile use cases; build image, dependency, and version management capabilities to support reproducible environments, resets, snapshots, and cloning.
  4. 04Improve agent tool calling, remote control, and session management, and establish mechanisms for log collection, resource management, and reclamation to keep execution environments isolated, secure, and observable.
  5. 05Stay current with open-source and cloud sandbox solutions, and work with algorithm, data, and product teams to turn new task environments, agent tools, and evaluation requirements into reusable platform capabilities.

What we look for

  1. 01At least 1 year of experience in backend, infrastructure, distributed systems, or cloud platform development, including building, deploying, and maintaining these systems.
  2. 02Proficiency in at least one programming language, such as Python, Go, Rust, or C++, with strong engineering design, testing, and troubleshooting skills.
  3. 03Proficiency with AI coding tools such as Claude Code or Codex for daily development, using codebase context to break down requirements, implement code, test, debug, and refactor, and ensuring quality through code review and validation.
  4. 04A solid understanding of operating systems, processes, networking, and storage fundamentals, with hands-on experience in at least one of the following: containers and virtualization, cluster scheduling, remote execution, or device management.
  5. 05An understanding of state management, idempotency, timeouts, retries, and failure recovery in asynchronous tasks and distributed systems, with the ability to analyze and resolve stability and performance issues across the execution pipeline.
  6. 06Ability to research technical solutions and validate prototypes for rapidly changing business needs, assess their suitability, limitations, and engineering costs, and drive implementation.

Nice to have

  • Experience with agent sandboxes, cloud development environments, browser clusters, or CI execution platforms; experience researching, using, or contributing to E2B, OpenSandbox, or similar solutions.
  • Development or hands-on experience with container runtimes, KVM / QEMU, Firecracker, Kata Containers, or gVisor.
  • Experience developing desktop or mobile execution environments, remote control systems, device clusters, emulators, or virtualization platforms.
  • Experience integrating, extending, or developing AI coding tools or agent harnesses, with an understanding of context management, tool calling, permission controls, session recovery, skills, and multi-agent collaboration.
  • Experience with agent benchmarks, automated evaluation, batch rollouts, or training data production platforms, and familiarity with large model post-training, including SFT and RLVR.
Apply for this role