← Back

Careers

Build configurable AI employees.

Tenor builds configurable AI employees with tools, memory, permissions, and working context, and embeds them inside the companies they serve.

Open roles
Before applying

Start with how we work.

Every role here comes with the same trade: high agency, careful speed, systems ownership, and evidence over theater. Before the role description, we show a short note on that bar so candidates can decide whether the environment is actually for them.

For more detail, read what working here is like and how our interview process works.

How we work

Read this before the role description.

This is not a normal software job. Tenor is building the infrastructure for configurable AI employees that do useful work inside real companies. The bar is not a demo or a clever agent loop. The bar is whether the system keeps working when the role is customer-defined, bounded, stateful, permissioned, and live.

01 · How we work
  • We move quickly, but only trust progress that survives inspection.
  • We expect people to own systems end to end: behavior, state, failures, fixes, and evidence.
  • We use coding agents by default. The job is to direct them well, judge their output, and verify what changed.
  • We prefer direct arguments, clear tradeoffs, and shipped improvements over long alignment rituals.
02 · Who tends to thrive

People do well here when they are comfortable with messy systems, allergic to fake progress, and motivated by infrastructure that compounds across many employees and environments.

If you want a fully mapped job, this will feel too early. If you want to define the infrastructure for configurable AI employees, keep reading.

03 · What we need

A world-class team can do almost anything. What we need are people willing to face the hard parts with us: relentless, direct, and unusually hard to discourage.

04 · This is not for you if
  • You want a narrow role with clean boundaries and a manager translating ambiguity into tickets.
  • You are excited by agent demos, but not by the reliability work that makes them survive real users.
  • You prefer looking productive to proving the system actually changed.
  • You use AI tools as a way to avoid understanding the code, state, logs, or failure mode yourself.
  • You are uncomfortable owning a live surface when the work spans product, infrastructure, and operations.
To the role →