We hire engineers
we’d trust with our own product.

Every candidate goes through a CV screen, a live interview and a one-day task that we build and run ourselves. In our last hiring round, 5 out of 155 candidates passed.

Four steps.
5 of 155 passed.

Our last hiring round, August to October 2026.

155candidates
~30interviews
15tasks
5passed
  1. 155candidates

    Applied and screened

    Every CV is read against the role: what the person has proven, what they only claim, and the level it points to.

  2. ~30interviews

    Live interview

    One hour, in person, with a lead engineer. Questions are written for this candidate’s CV.

  3. 15tasks

    One-day task

    A small real product: a Flutter app from a Figma file, a React admin screen, a NestJS API, or a QA bug hunt.

  4. 5passed

    Passed

    We build and run every submission, read its git history and check how honestly the candidate judged their own work.

Screened with HR Drone,
decided by people.

HR Drone, our own AI recruiting product, runs the first passes. It reads each CV, sorts every claim into proven or only stated, and places the candidate on our level scale from Intern to Senior 3. Then it drafts an interview built to test that reading.

Reads for evidence

A CV line counts only when the work behind it can be checked.

Blind to background

Name, photo, age, gender, nationality and university are ignored.

Humans decide

The AI writes a recommendation. A lead engineer runs every interview and the Head of Engineering makes every hire.

A task with
a hidden answer key.

Each task has known gaps planted in it. A junior usually finds about two, a mid-level engineer three or four, a senior five or six. We count what they find, fix and admit.

01

We run it

Every submission is built and used, with screenshots of what breaks.

02

“What is still wrong with this?”

Every README must answer it. Honest answers score higher than polished ones.

03

AI is allowed

Candidates may use any AI tool. We test whether they understand what they shipped, and we give their written project brief to a fresh AI agent to see if it can continue the work.

04

Two scores, never averaged

Technical skill and judgment are scored separately. Strong opinions with weak code is still a no.

05

Questions are welcome

Asking good questions during the task counts in the candidate’s favour.

Hired for
the long run.

We hire people who share our values and want to grow with us. That’s why the engineers who start your product are still there when it scales.

M-One team members working togetherM-One engineers in the Yerevan office
50+engineers in-house
1-11 Specialist – 1 Product
0freelancers
4+ yrsaverage time with us

English classes in the office, health insurance for everyone, and team leads whose job includes mentoring.

Tell me what you’re trying to build.

Thirty minutes with me, not a salesperson. A straight answer on whether we can help, and a first view on how.