Flow 07 of 08

AI Recruitment Engine

Resume parsing, skill extraction, education verification, weighted scoring, and intelligent candidate ranking.

Story Line

The AI Recruitment Engine is the brain of the platform. It receives two inputs — the resume file (PDF, DOCX, or scanned image) and the questionnaire answers — and produces one output: a ranked, scored, recommendable candidate profile.

First, the resume is parsed. If it is a digital PDF or DOCX, a structured parser reads the text directly. If it is a scanned image, an OCR engine reconstructs the text first. Sections are detected (contact, experience, education, skills, certifications) and each is normalised into a canonical schema.

Next, skills are extracted against a taxonomy and matched to the job description. Years of experience are computed, education is verified against an institution dictionary, and certifications are validated. Each dimension contributes a sub-score: skills (40%), experience (30%), education (15%), questionnaire fit (15%).

The engine produces a weighted final score, generates a human-readable AI summary ("Strong match on CAD & FEA, mid-tier on team-leadership, no certification gaps"), and pushes the candidate onto the ranked list inside the recruiter's dashboard. The same engine periodically re-ranks the talent pool when new jobs are posted.

Workflow Diagram

%%{init: {'theme':'base','themeVariables':{'primaryColor':'#FFF1E6','primaryTextColor':'#1a1a1a','primaryBorderColor':'#F77020','lineColor':'#F77020','secondaryColor':'#FFE4D1','tertiaryColor':'#ffffff','clusterBkg':'#FFF8F2','clusterBorder':'#F77020','fontFamily':'Inter','fontSize':'14px'},'flowchart':{'curve':'basis','htmlLabels':true}}}%% flowchart TD classDef trust fill:#F77020,stroke:#c4541a,stroke-width:2px,color:#ffffff classDef action fill:#FFF1E6,stroke:#F77020,stroke-width:2px,color:#1a1a1a classDef ai fill:#FFE4D1,stroke:#F77020,stroke-width:2px,color:#7a3a10 classDef user fill:#ffffff,stroke:#F77020,stroke-width:2px,color:#1a1a1a classDef success fill:#F77020,stroke:#9c4214,stroke-width:3px,color:#ffffff classDef decision fill:#ffffff,stroke:#F77020,stroke-width:2px,stroke-dasharray:4 2,color:#1a1a1a Start([Application Submitted]) --> Inputs[Collect Inputs] Inputs --> Resume[Resume File] Inputs --> Answers[Questionnaire Answers] Inputs --> JD[Job Description & Weights] Resume --> Format{File Format?} Format -- PDF --> ParsePDF[PDF Text Extractor] Format -- DOCX --> ParseDOCX[DOCX Parser] Format -- Image / Scanned --> OCR[OCR Engine] OCR --> ParsedText[Reconstructed Text] ParsePDF --> ParsedText ParseDOCX --> ParsedText ParsedText --> Sectionise[Detect Sections] Sectionise --> Contact[Contact Block] Sectionise --> ExpSec[Experience Block] Sectionise --> EduSec[Education Block] Sectionise --> SkillsSec[Skills Block] Sectionise --> CertSec[Certifications Block] SkillsSec --> SkillNorm[Normalise to Skill Taxonomy] SkillNorm --> SkillMatch[Match Skills vs JD] SkillMatch --> SkillScore[Skills Sub-Score 40 percent] ExpSec --> ExpYears[Compute Years of Experience] ExpYears --> ExpRel[Relevance to JD] ExpRel --> ExpScore[Experience Sub-Score 30 percent] EduSec --> EduVerify[Verify Institution & Degree] EduVerify --> EduScore[Education Sub-Score 15 percent] Answers --> AnsClean[Normalise Answers] AnsClean --> AnsFit[Compute Fit vs Configured Weights] AnsFit --> AnsScore[Questionnaire Sub-Score 15 percent] SkillScore --> Combine[Combine Weighted Scores] ExpScore --> Combine EduScore --> Combine AnsScore --> Combine Combine --> Final[Final Match Score 0 to 100] Final --> Threshold{Above Threshold?} Threshold -- No --> LowFlag[Flag Below-Threshold] LowFlag --> Pool[Send to Talent Pool] Threshold -- Yes --> Summary[Generate AI Summary] Summary --> Strengths[Strengths & Highlights] Strengths --> Gaps[Gaps & Risks] Gaps --> Profile[Build Comprehensive Profile] Profile --> Ranking[Rank Within Job Pipeline] Ranking --> Recommend[Surface Top-N Recommendations] Recommend --> Pipeline[Push to ATS Shortlisted Stage] Pipeline --> HRView([Visible on Recruiter Dashboard]) Pool --> Periodic[Periodic Re-Rank Job] Periodic --> NewJobs{New Jobs Posted?} NewJobs -- Yes --> Combine NewJobs -- No --> Idle([Idle — Await Trigger]) class Start,Inputs,Resume,Answers,JD,Contact,ExpSec,EduSec,SkillsSec,CertSec user class Format,Threshold,NewJobs decision class ParsePDF,ParseDOCX,OCR,ParsedText,Sectionise,SkillNorm,SkillMatch,ExpYears,ExpRel,EduVerify,AnsClean,AnsFit,Combine,Final,Summary,Strengths,Gaps,Profile,Ranking,Recommend,SkillScore,ExpScore,EduScore,AnsScore,Periodic ai class Pipeline action class LowFlag,Pool trust class HRView success class Idle action
Resume Parsing

PDF, DOCX, and scanned image fallback via OCR for full coverage.

Skill Taxonomy

Skills normalised to a canonical taxonomy and matched against the JD.

Weighted Scoring

Skills 40% · Experience 30% · Education 15% · Questionnaire 15%.

Live Ranking

Top-N candidates surfaced and re-ranked as new jobs are posted.