Skip to content
Clexerl
All roles

Frontier AI · Agents Platform

Software Engineer, Sandboxing & Code Execution

Give agents safe, fast computers to work in: isolation, snapshots and secure egress.

  • San Francisco · Remote (US)
  • Remote-friendly
  • Full-time
  • $280K – $460K

About the role

Agents that write and run code need somewhere safe to do it. The Agents Platform team builds those computers: isolated, fast to start, cheap to hold, and locked down so that whatever the model runs cannot escape, exfiltrate data or reach systems it should not.

Clexerl Go agents already coordinate scheduling, tasks and CRM work through dense tool use, and increasingly they run code to analyse data, transform files and test their own changes. Frontier AI also uses the same environments for reinforcement learning, where millions of short episodes need clean, resettable machines. Isolation quality is both a product feature and a safety control.

We assume every workload is hostile. Designs go through a threat model before they ship, and we run adversarial tests against our own boundaries.

What you’ll do

  • Build the isolation layer for code execution using microVMs, user-space kernels or container hardening, and choose between them on evidence.
  • Implement snapshot and restore so sandboxes start in milliseconds and agents can fork, checkpoint and roll back state.
  • Design secure egress: allow-listed networking, DNS controls, credential injection that never exposes secrets to model-generated code.
  • Manage resource limits for CPU, memory, disk and process counts, and detect abuse such as fork bombs or cryptomining.
  • Build the control plane that schedules sandboxes at high churn, with warm pools and predictable cost.
  • Work with safeguards and security teams to threat-model each new capability before release.

What you’ll bring

  • Strong systems programming in Rust, Go or C, with working knowledge of Linux namespaces, cgroups, seccomp and capabilities.
  • Experience with virtualisation or sandbox technologies, such as KVM-based microVMs or syscall interception.
  • Understanding of container escape classes and how to reason about attack surface, not just configuration.
  • Familiarity with filesystem snapshotting, copy-on-write storage and memory snapshot techniques.
  • Experience building control planes or schedulers that handle many short-lived workloads reliably.
  • A security mindset applied to ordinary engineering decisions.

Nice to have

  • Prior work on userfaultfd-based lazy restore or similar fast-start techniques.
  • Experience with eBPF for observability or enforcement.
  • Background in vulnerability research or red-team exercises against isolation boundaries.
  • Familiarity with reinforcement learning environment infrastructure.

Work you might take on

  1. 01

    Cut sandbox cold start to tens of milliseconds by restoring pre-warmed memory snapshots with lazy page loading.

  2. 02

    Build an egress proxy that injects short-lived credentials per request so model-generated code never sees a long-lived secret.

  3. 03

    Add fork-and-rollback so an agent can branch a running environment, try two approaches and keep the better result.

  4. 04

    Write an adversarial test harness that runs known escape techniques against every sandbox build and fails the release on any breakout.

Compensation

$280K – $460K

The US base salary range for this role. Every offer also includes equity in Clexerl and, when it spins out, in the venture you help build. We tell you the reasoning behind your number.

Benefits

  • Health, fully covered

    Medical, dental and vision for you and your dependants, premiums paid in full.

  • Two kinds of equity

    Equity in Clexerl, plus a stake in the venture you help build when it spins out.

  • 22 weeks parental leave

    Fully paid, for every parent, however your family arrives.

  • Rest that sticks

    Flexible time off with a four-week minimum we actually check on, plus a company-wide week in December.

  • Learning and compute

    A yearly budget for courses, conferences and books, and compute credits for side projects.

  • 401(k) with a 4% match

    Matched from your first paycheque and vested immediately.

How we hire

  1. Apply 5 minutes

    Send links to what you have built and a few lines on why this role. No cover letter needed.

  2. Intro call 30 minutes

    A conversation with the hiring manager about the work, the team and what you want next.

  3. Work sample 2–4 hours

    A paid take-home or a pairing session on a real-shaped problem. You pick the format.

  4. Studio day One day

    Meet the team, go deep on your work and ours. Remote or in person.

  5. Offer Within 5 days

    A written offer with the full range, the equity and the reasoning behind both.

Apply for this role

Opens an email with a short template. Links beat résumés.

All Frontier AI roles