Resume Radar

Home page
Result
About
Query
Result
Result
Result

Inspiration

Job seekers often get rejected without knowing why.

Most resume tools rewrite content, but none quantify skill gaps against real job requirements.

We built Resume Radar to turn resume screening from subjective feedback into measurable career intelligence.

What it does

Resume Radar analyzes a candidate’s CV against a job description and generates:

Overall match score (%)
Confidence score (0-100)
Transparent score breakdown (skills / experience / seniority / domain)
Skill gap breakdown
Missing critical competencies
Gap table (Skill | Required | Candidate | Gap)
Market demand insights (with source label/date)
Estimated salary band + market value uplift (currently static sample market dataset)
Personalized 6-week improvement roadmap

It transforms resume optimization into a data-driven dashboard experience.

Instead of guessing, users see exactly:

What they’re missing.

How much it matters.

What to do next.

How we built it

Frontend dashboard in React + Vite (with lazy-loaded chart components)
Backend analysis API (POST /api/analyze) with health endpoint (GET /health)
Server-side Gemini extraction for resume/JD parsing (no frontend API key usage)
Rule-based parsing fallback for resume and JD when LLM extraction is missing/weak
Deterministic weighted match scoring algorithm (transparent subscores)
Skill normalization via canonical dictionary + synonym mapping (rule-based)
Static sample market dataset for demand trends / salary estimate / uplift (explicitly labeled in UI)
Interactive analytics dashboard with gap table, radar chart, market trends chart, and 6-week roadmap

The system combines NLP extraction, scoring logic, and visual intelligence.

Challenges we ran into

Handling inconsistent resume formats
Supporting scanned/image-only PDFs (OCR fallback not fully integrated yet)
Normalizing synonymous skills (e.g., React vs React.js)
Preventing LLM hallucination in skill extraction
Designing transparent scoring logic while preserving useful UX
Making the app resilient when backend API or LLM extraction is unavailable

We mitigated these using schema validation, canonical skill dictionaries, deterministic scoring, rule-based extraction fallback, and backend health checks surfaced in the UI.

Accomplishments that we're proud of

Built a working end-to-end resume-to-dashboard pipeline with a backend API
Moved Gemini usage server-side and kept the frontend secret-free
Implemented explainable deterministic match scoring + confidence score
Added rule-based fallback parsing so analysis still works when LLM extraction fails
Designed an intuitive gap table + chart-based visualization system
Delivered structured 6-week improvement roadmaps with interactive tracking