Staff Engineer · Agent runtime

About the role

You will own the loop: the plan/execute/verify state machine that decides when Friday stops, when it re-plans, and how it orchestrates agent runs. Solo development today; you'd be the second engineer on the team.

The Work

• Own the agent loop end-to-end: state machine, scheduler, agent orchestrator, verifier.
• Help design the protocol between daemon and CLI/desktop/mobile clients.
• Drive the PTY sandbox primitives: process isolation, network policies, kill switches.
• Pair with the founder daily for the first three months. After that, divide ownership by area.
• Write the engineering posts that explain how the loop works — internally and publicly.

First Six Months

A real plan, not a wish. What success looks like, broken down by month:

• Month 1: Read every line of the existing daemon. Use Friday daily, list every grievance.
• Month 2–3: Design the v1.0 protocol with the founder. Land the new daemon scaffolding.
• Month 4–5: Cut the agent loop over to the new protocol. Migrate desktop and CLI clients.
• Month 6: Open the public beta wider — first 500 users, real feedback loop.

Requirements

• 8+ years in production systems engineering.
• Rust in production for at least 2 of those years. Comfortable in async Tokio.
• Shipped at least one agent loop, workflow runtime, or background-task system that real users relied on.
• Experience with PTY, IPC, or process supervision (systemd, supervisord, k8s pods, your choice).
• Written technical posts that other engineers read voluntarily.

Nice to Have

• TypeScript / SolidJS in clients (Tauri-friendly).
• Compiler / language tooling background.
• Prior open-source maintainership.

Compensation

Senior staff compensation, structured as runway. Equity in Friday as it forms.

Reporting

You will report directly to the founder. The team currently consists of you and the founder (software development).