Skip to main content
Run comprehensive AI evaluations to score and rank candidates based on your customized job criteria.

​
Prerequisites

Before running an evaluation:
You have created a job posting
You have added at least one candidate to the job
Your AI evaluation weights are configured (must total 100)

​
Run the evaluation pipeline

1

Select your job

Ensure the correct job is selected in the dropdown menu at the top of your dashboard.
2

Click Run AI Pipeline

In the Candidate Cohort panel, click the Run AI Pipeline button (located at the bottom).The button will be disabled if you have no candidates added to the job.
3

Wait for processing

The button will display “Processing via AI…” while the evaluation runs. The system:
  • Updates your job configuration
  • Evaluates each candidate through multiple AI agents
  • Calculates weighted scores
  • Performs fraud detection when needed
Large candidate pools may take several minutes to process. Do not navigate away from the page during evaluation.
4

View results

When complete, the results appear automatically in the right panel. Candidates are ranked by final score in descending order.

​
Understanding evaluation scores

The AI pipeline generates five component scores for each candidate:

​
Skill score

Measures how well the candidate’s skills match your required and preferred skills. The Decision Intelligence agent analyzes:
  • Verified skills from resume and projects
  • Primary programming languages
  • Project complexity indicators
  • Consistency between claimed and demonstrated skills

​
GitHub score

Evaluates the candidate’s GitHub profile activity and code quality. The GitHub Analyst agent examines:
  • Repository activity and contributions
  • Project quality and complexity
  • Code patterns and best practices
  • Open source participation
Candidates without GitHub links receive a score of 0. Encourage candidates to include their GitHub profiles for accurate evaluation.

​
Interview score

Assesses interview responses (if provided). The Interview Grader agent analyzes:
  • Answer quality and relevance
  • Technical depth
  • Communication clarity
  • Problem-solving approach
If no interview answers are provided, the system uses resume length as a fallback metric.

​
Experience score

Compares the candidate’s years of experience against your minimum requirement. The Decision Intelligence agent evaluates:
  • Total years of relevant experience
  • Experience level classification (junior, mid, senior)
  • Career progression patterns
  • Domain expertise depth

​
Integrity score

Measures profile consistency and detects potential fraud. The system:
  • Compares claims across different data sources
  • Flags inconsistencies between GitHub activity and interview responses
  • Detects resume padding or exaggeration
  • Identifies bias indicators in language
When GitHub and interview scores differ by more than 30 points, the Integrity Analyst agent is automatically triggered. This may apply penalties to the final score.

​
Final score calculation

The final score is a weighted average of the five component scores: 
Final Score = (Skill × W₁) + (GitHub × W₂) + (Interview × W₃) + (Experience × W₄) + (Integrity × W₅)
Where W₁ through W₅ are your configured evaluation weights (totaling 100%). If fraud is detected, an integrity penalty is subtracted from the final score.

​
Review evaluation results

​
Results table

The AI Ranking Engine Results table displays candidates with the following columns:
  • Rank - Position in the leaderboard (#1 = highest score)
  • Candidate Profile - Name (or anonymized ID in blind mode)
  • Final Score - Overall weighted score (0-100)
  • Skill, GitHub, Interview, Integrity - Individual component scores (clickable to sort)
  • Risk Flag - Low Risk (green), Medium Risk (yellow), or High Risk (red)
  • Verdict - Strong Hire, Hire, Waitlist, or Consider
  • Action - “View Details” button (appears on hover)
Click any column header with the ↕ symbol to sort candidates by that metric.

​
Visual analytics

The dashboard displays several charts to help you analyze your candidate pool: Final Score Leaderboard - Horizontal bar chart showing top 10 candidates by final score. The #1 candidate is highlighted in green. Risk Profile Distribution - Pie chart breaking down candidates by risk level (Low, Medium, High). Skill Match Breakdown - Bar chart comparing skill scores across top candidates. Average Profile - Radar chart showing the average scores across all five evaluation dimensions. GitHub Activity - Bar chart highlighting candidates with strong GitHub profiles.

​
Candidate details

Click View Details on any candidate row to see:
  • Individual radar chart of their five scores
  • Detailed strengths list (verified skills, reasoning)
  • Weaknesses and risk factors
  • Fairness adjustment notes (if applicable)
  • Fraud investigation results (if triggered)

​
Re-run evaluations

You can re-evaluate candidates at any time:
  1. Update your job requirements or evaluation weights
  2. Click Run AI Pipeline again
  3. Previous evaluation results are automatically replaced
Re-running evaluations recalculates all scores from scratch. This is useful after updating job criteria or adding interview data.

​
Next steps

Blind review mode

Review candidates anonymously to reduce bias

Creating jobs

Create another job posting

Build docs developers (and LLMs) love

Get started for freeTalk to us