Vetting process
Only the top 1% make it to your codebase
Job boards optimize for volume. We optimize for proof — live sessions, references, and a paid trial before anyone joins your team. Read why clients choose us or browse vetted profiles.
Acceptance rate
Top 1%
accepted after full vetting pipeline
Four gates — every role
- 01
Application screening
We review experience, shipped work, and role fit. The majority of applicants stop here — we only advance candidates with verifiable production experience.
- 02
Live technical assessment
Developers complete a live coding session on problems relevant to their specialty. Designers present portfolio work and walk through process. AI tools may be used; we evaluate whether they understand what ships.
- 03
Reference and communication check
We talk to past clients and managers. Clear updates, timezone reliability, and ownership matter as much as raw technical skill.
- 04
Paid trial week
Before a full placement, every hire completes a paid trial on a scoped task — yours or a representative brief. You see real output, not interview performance.
AI-native vetting
We vet engineers who use Cursor and Claude — and know what ships
AI-assisted coding is allowed in our live technical review. We do not reward tab-complete speed. We score whether candidates understand the diff, write tests, and can defend merges to main.
- Tools in the assessment
Cursor, Claude, and Copilot are fine — same as your team probably uses. Output still needs review, edge-case handling, and a clear explanation of tradeoffs.
- Production proof
We prioritize candidates who have shipped LLM features with billing, auth, and observability — not notebook demos. Magicshot.ai is the benchmark we cite with clients.
- Paid trial on your repo
Before a long placement, every hire completes a scoped trial in your codebase or a representative brief. You see real PRs, not interview theater.
After placement
Vetting does not end at day one. We run check-ins with you and the hire. If a vetted developer does not perform to the standard we promised, we replace them at no extra cost. That guarantee is in writing — Patrick Luckett cited it as the reason their arrangement was a clear yes.
Vetting FAQ
What is the acceptance rate?
We accept roughly the top 1% of applicants. Most candidates are rejected before a live technical review. Only those who pass coding or portfolio review, references, and a paid trial week are placed with clients.
What is tested in the live coding session?
Real problems relevant to the role — building a small feature, debugging existing code, or walking through system design. For designers, a portfolio review and a scoped design task. AI tools may be used; understanding and review of output is evaluated.
What happens during the paid trial week?
The candidate works on a scoped task on your project or a representative brief. You see output before committing long-term. If it is not the right fit, you stop — no penalty.
How is performance monitored after placement?
We run check-ins with you and the hire. If performance drops below the standard we promised, we replace them at no extra cost under our nonperformance guarantee.
Do you vet for communication and timezone overlap?
Yes. Reference checks cover async communication and clarity. We match timezone overlap to your working hours — typically 4–6 hours minimum for US clients.
Do your developers use Cursor or GitHub Copilot?
Many do — daily, in production repos. Our live vetting allows AI tools; we evaluate whether candidates review output, write tests, and understand what merges. Browse profiles filtered by tool on the developer directory.
Can hires work inside our Claude project or Anthropic org?
Yes. We place Claude engineers who have integrated the Anthropic API, tool use, and RAG in live products. Tell us your setup on the book-a-call form and we match for that stack.
See the vetting process in action
Book a call. We will walk through how we match for your stack and quote a realistic start date.