Agent Pipeline

All 9 agents

Phase 1 — Evaluation Pipeline

Candidate Submission

Name · Email · Resume · GitHub · LeetCode · Credly · Job ID

01

Planner Agent

Parse & Structure

Parses raw form data into a clean JSON envelope — resume, metadata, and job description combined.

02

Screener Agent

Resume × JD Fit

Compares resume against job description. Returns skill match score, experience gaps, and red flags.

03

GitHub Agent

Repository Analysis

Analyzes top languages, repo count, activity recency, and coding history depth against role requirements.

04

LeetCode Agent

DSA Proficiency

Evaluates solve counts (Easy/Medium/Hard), contest rating, global rank, badges, and top DSA topic clusters.

05

Credly Agent

Certification Verification

Checks Credly for issuer-verified badges. Also reads resume for NPTEL, Udemy, and Coursera certifications.

06

Analyzer Agent

360° Candidate Profile

Synthesizes all 4 analysis blocks into combined strengths, weaknesses, red flags, and a weighted overall score.

07

Decision Agent

APPROVE / REJECT

Issues final recommendation with confidence level, HR reasoning, and generates personalised email and Slack message drafts.

Pipeline Output — Firestore

overall_score87 / 100
decisionPENDING_HR
resume_score82 / 100
github_score90 / 100
leetcode_score78 / 100
combined_strengths[…3 items]
combined_red_flags[…1 item]
hr_decisionPENDING
HR reviews in portal

Phase 2 — Post-Decision Agents

HR Approves →
08

Selection Agent

Approval Flow

Writes a warm congratulations email with next-step details and pings the HR Slack channel.

Approval email→ candidate via Resend
Slack ping→ HR channel
hr_decision → APPROVED
HR Rejects →
09

Rejection Agent

Rejection Flow

Writes an empathetic rejection referencing the candidate's actual gaps — never a generic template.

Rejection email→ candidate via Resend
Slack ping→ HR channel
hr_decision → REJECTED