Systems Engineer, Hypervisor and Kernel

Workers IO builds deterministic simulation environments for verifying mission-critical software.

Modern systems fail in ways no test suite catches: a retry lands during a partial outage, two writes interleave, a dependency times out at exactly the wrong moment. Teams find these bugs in production, after an incident, with no way to reproduce them.

We fix that. Workers IO runs your entire system inside deterministic virtual machines and searches through millions of interleaving, faults, and timings to find the exact conditions that break the promises you care about: no lost writes, no double fills, no split-brain.

Coding agents are writing more of the world's software, and they need a feedback loop stronger than "the tests passed." We give agents and engineers a way to verify a change against production reality.

Teams where 99.99% uptime is table stakes and someone is accountable for incidents: payments, trading, infrastructure, healthcare, and the platforms other companies run on.

• Backed by Y Combinator and the Solo Founders Program, and some of the best angels.
• Small founding team, deep systems focus, high ownership

If you like operating systems, distributed systems, formal-ish thinking applied to real software, and you want your work to be the reason a payment system did not lose a write, we should talk.

What we are looking for

• Deep experience with hypervisors, virtual machine monitors, or emulators
• Real work inside the Linux kernel: schedulers, timers, interrupts, drivers, or memory management
• Comfort in C and Rust, and comfort reading code you did not write
• A habit of measuring before guessing, and of writing down what you learned

Nice to have

• Contributions to Linux Kernel, KVM, QEMU, Firecracker, or similar
• Experience with record and replay, deterministic execution, or simulation testing
• Time spent running production systems on bare metal

• Systems: Rust, C, Linux, custom hypervisor, bare-metal AWS
• Control plane: Convex (TypeScript), WorkOS, GitHub integration
• Engineering practice: Deterministic simulation testing of our own platform, strict quality gates in CI (coverage floors, mutation testing, ratcheted lint budgets)

• Extending deterministic control to new device classes and guest subsystems
• Shrinking failing traces to minimal reproductions automatically
• Exploring schedule and fault spaces efficiently across millions of seeds
• Making snapshot restore faster and cheaper on metal
• Giving coding agents a tight, trustworthy verification loop on every change
• Turning kernel-level evidence into a clear story an engineer can read in a minute

People who have shipped a kernel patch, debugged a hypervisor, built a distributed database, or spent a weekend making something reproducible that had no business being reproducible. You will own problems end to end, from the host down to the guest and back up to the UI that shows the result.