The right AI engineer. Not just a resume.
Caliber finds and places AI and ML engineers at fast-growing startups. Every candidate is assessed by working AI engineers. You get a written scorecard with each brief, not a pile of CVs.
The problem
Everyone claims to know AI. Most don't.
Every engineer has “LLM” and “RAG” on their resume after a weekend with the OpenAI API. Most recruiters cannot tell the difference between someone who shipped a RAG pipeline to production and someone who followed a tutorial. You end up finding this out four rounds in.
Most recruiters
- See "LLM" on a resume and send it over
- Can't tell RAG from calling an API
- Send 15 profiles, hope something sticks
- Get paid either way
Caliber
- Asks: are you building RAG, fine-tuning, or infra?
- Every candidate screened by a working AI engineer
- Three profiles per role, each worth your time
- Paid only when someone joins
Why Caliber
Different from every recruiter you've tried before.
Domain specialists
Only AI.
We work in AI and ML engineering and nothing else. Our network, our questions, and our judgment are all focused on this one space.
Engineer-assessed
Screened by engineers, not HR.
Every candidate does a written technical assessment, then a call with a working AI engineer. You get a scorecard with each brief: what we asked, what they said, and where they're weak.
Aligned incentives
You pay when they join.
12% of CTC, paid on the date they start. If they leave in the first 90 days, we find a replacement at no charge.
The process
Three steps. No wasted rounds.
We learn your team.
A 30-minute call. We ask what you're building, what the team looks like, and what went wrong with past hires. We send back a written role spec for you to confirm before we start.
Role scoping · Technical bar · Team context
We assess candidates.
Written technical assessment tailored to the role: RAG systems, model training, MLOps, inference infra each get different questions. Then a call with an AI engineer. You get a scorecard for each candidate we submit.
Written assessment · Engineer review · Scorecard
See a sampleYou talk to three people.
We send a maximum of three candidates per role. Each one has passed our technical screen and knows your role and company before they pick up the phone. Your team's time goes on people worth talking to.
Max 3 per role · Pre-qualified · Offer support included
Roles we place
If it ships AI, we can find it.
High demand right now
Who we are
Engineers who have
hired engineers.
Caliber is built by engineers who have shipped AI systems in production and hired the people who built them.
We screen the way we would want to be screened: on what someone actually built, the calls they made under pressure, and whether they can walk through a hard tradeoff and tell you what they got wrong. Not how many times a keyword shows up.
How we assess
Real work, not prep work
We ask about systems they built, not problems they practiced.
Different questions per role
RAG engineers, model trainers, and MLOps get assessed differently.
Scorecard with every candidate
You see what we asked, what they said, and where they struggled.
Work with us
Tell us about the role.
We respond within one business day. If it's not a fit for us, we'll say so rather than string you along.
No exclusivity required
Keep working with your existing team and recruiters in parallel.
90-day replacement guarantee
If the hire leaves in the first 90 days, we find a replacement at no charge.
Contingency-based
12% of CTC. You pay when they join their first day.