Talent Navigator
Vetting system

Five stages between an application and your inbox

Candidates prove their skills rather than describe them. By the time a profile reaches you it has passed a technical test taken on camera, a review of how they actually write code, and a check on their education and work history. The full report comes attached, and reading it costs you nothing.

Open a real sample reportBrowse vetted talent

Out of every 100 who apply

Applied
Everyone who applies
100%
Screening
Clear application screening
15%
Technical
Pass the technical assessment
7%
Analysis
Survive the analysis pass
6%
Proctored
Complete a proctored test cleanly
5%
The process

What each stage actually tests

No stage is a formality. The rejection rate next to each one is the share of everyone who reaches it and does not get past it.

01

Application screening

85% rejected

Resume quality, portfolio, GitHub, work history and communication samples.

  • Minimum 3 years of professional experience
  • Strong portfolio with production-level projects
  • Education and certification verification
  • Initial English proficiency check
  • Profile and work history review
02

Technical assessment

55% rejected

Domain-specific problems over 30 to 90 minutes, language-specific, built on real project scenarios.

  • Three timed coding challenges, easy through hard
  • System design and architecture questions
  • Language-specific proficiency tests
  • Code quality and best practice evaluation
  • Problem-solving approach analysis
03

AI-powered analysis

40% rejected

Pattern detection, code quality analysis, consistency checks and behavioural signals.

  • Code pattern and style analysis
  • Algorithmic efficiency scoring
  • Technology stack proficiency mapping
  • Learning curve prediction
  • Team fit compatibility scoring
04

Supervised final test

30% rejected

Taken on camera with the browser locked, so we know the person who scored is the person you hire.

  • 45 minute test, supervised end to end
  • Webcam snapshot verification
  • Tab switching detection
  • Copy and paste blocking
  • Behavioural analysis
05

Elite talent pool

Top 5% only

Verified, ready and available, with a live profile you can search.

  • Final comprehensive review
  • Communication skills assessment
  • Availability and timezone confirmation
  • Rate alignment and onboarding prep
  • Profile creation and talent pool entry

And then you see the report

Not a badge. Not a star rating. The full assessment, section by section, with the integrity check attached. Here is a real one.

Open the live report
A real report

This is what you read before you spend a credit

Every figure below is taken from a live report on our assessment platform. Nothing is smoothed over: one section scored 40 percent and one tab switch was flagged.

Strong hireHigh confidence

Senior AI / ML Engineer

18 of 21 questions correct · 42 minutes · 2 minutes average per question

86of 100

Score by section

Multiple choice130/130100%
Coding challenge35/35100%
System design13/1587%
Fill in the blanks20/5040%

Score by skill

Python and ML frameworks100%
AWS for ML workloads100%
LLMs and RAG75%
MLOps67%

Strengths

  • Excellent Python programming and ML frameworks, TensorFlow and PyTorch
  • Strong understanding of LLMs and RAG architectures
  • Solid MLOps practice with Docker and Kubernetes
  • Good AWS experience for ML workloads across SageMaker, Lambda and S3
  • Clear problem-solving approach with clean, efficient code

Where they are weaker

  • Could improve understanding of advanced optimisation techniques
  • Room for growth in large-scale distributed training

We publish the gaps as well as the strengths. A report that only ever flatters the candidate is not evidence, it is a brochure.

Integrity check

Proctoring violations
1
Warnings issued
1
Tab switches
1, flagged
Time consistency
97%
Clean test
Yes

Suggested interview questions

  1. 01Describe your experience building RAG systems at scale.
  2. 02How do you approach model versioning and experiment tracking?
  3. 03Walk us through your LLM fine-tuning process.
LLM chatbot development and fine-tuningRAG system implementation with vector databasesML pipeline automation with SageMakerProduction model deployment on Kubernetes

Downloading the full report costs nothing, on every profile.

Open this report in full
Anti-cheating

A score is worthless if the test was gamed

Assessments run under proctoring, and the integrity result travels with the report. In the sample above the candidate switched tabs once. We did not hide it. We scored it, flagged it, and still recommended the hire, because one flag on a 97 percent integrity score is a person checking the time.

See scored profilesCompare us to agencies
A candidate working through a timed assessment
45 minutes, on camera, browser locked. Every score on this site came from a session like this one.

Webcam monitoring

Face detection, identity verification and movement analysis

Tab control

Tab switch alerts, copy and paste disabled, browser lock

Keystroke analysis

Typing pattern verification and AI behaviour detection

Time analysis

Question timing consistency and rushed answer flagging

Why it holds up

Scores you can actually rely on

Verified skills, not claims

Every score traces back to work the candidate produced under observation.

Watched the whole way through

Camera, tab switching and typing are all monitored while the test runs, and anything unusual is flagged on the report.

Consistent scoring

Every candidate for a role answers the same questions and is marked the same way, so the scores mean something when you compare two people.

What vetting does not do

  • It does not replace your interview. It replaces the first three of them.
  • It does not measure how someone will fit your team. Only you can judge that.
  • It does not guarantee an outcome, which is why every placement carries a two week risk-free trial.
Browse vetted talentStart hiring free

Hiring abroad? We can employ them on our payroll so you do not need an entity in their country.