How Playmakers vets engineers.
Everyone on a Playmakers shortlist got there the same way: by doing real work, scored the same way as everyone else's. This is the whole round.
The round, start to finish.
- They apply to a real role.Every candidate applies to a role that is open now, and answers its questions.
- They build against the clock.The assignment is a timed, practical build (the role’s brief sets the window, usually a day): a real, live product, rebuilt with their own design. They can use any AI tools they like, and they commit their coding-agent session logs to the repository as they go.
- The work is scored automatically.Four scores, each out of 100: the session logs, the architecture, the walkthrough video and their GitHub activity.
- Below the bar is declined automatically.A submission that scores under the bar is declined without waiting on anyone. Everything above it is read by a person.
- A person picks who you meet.Above the bar, somebody on our team reads the work, invites the strongest to an interview, and chooses who goes on your shortlist.
- You get a shortlist of about five.Each person with their work attached: the live build, the code, the sessions, the walkthrough and the four scores. You decide after one call.
What the four scores mean.
Each is out of 100, and every one is on the shortlist you receive.
- Coding-agent logsHow they direct Claude Code, from the sessions they committed.
- ArchitectureIs the backend real, and does it hold up?
- WalkthroughCan they explain what they built, camera on?
- GitHub contributionsTheir public activity over the last year.
What clients ask about it.
Yes, any tools they like. How well they direct them is part of what we score, which is why they commit their coding-agent sessions.
Everyone given a brief rebuilds the same live product, so their work can be compared side by side, against something that already works.
Yes. The scores only decline what falls under the bar. Above it, somebody on our team reads the work and chooses who goes on your shortlist.
Their four scores, the live build, the source code, their Claude Code sessions and the walkthrough video, on one card.