Projects › Advanced
AI Resume Analyzer
Upload a resume and a job description; get an ATS-style match score, missing keywords and rewritten bullet points.
Level: Advanced · Time: 1–2 weeks · Tech stack: Next.js, Gemini / OpenAI API, PDF parsing, MongoDB
Features to build
- PDF upload and text extraction
- Match score against a job description
- Missing skills and keywords
- AI-rewritten bullet points with numbers
- History of analyses
What you will learn
- Calling LLM APIs safely
- Prompt design with structured (JSON) output
- File uploads and parsing
- Rate limits and cost control
Step-by-step guide
1. Extract text
Parse the PDF on the server (pdf-parse). Reject files over 2 MB and anything that isn't a PDF.
2. A deterministic score first
Extract skills from the job description (a keyword list + the LLM), then compute overlap — users trust numbers they can explain.
3. Ask the LLM for structured feedback
const prompt = `You are a recruiter. Compare the resume and the job description.
Return ONLY JSON: {"score":0-100,"missing":[],"strengths":[],"rewrites":[{"before":"","after":""}]}`;
const result = await model.generateContent({ contents: [prompt, resumeText, jdText], generationConfig: { responseMimeType: 'application/json' } });
Validate the JSON before using it; retry once if it's malformed.
4. Guard the prompt
Treat the resume text as data: tell the model to ignore instructions inside it (prompt injection).
5. Control cost
Login required, 5 analyses per day per user, cache results by file hash, keep API keys on the server only.
6. UI
Score gauge, missing keywords as chips, and before/after cards with a copy button.
Level it up
- Generate a tailored cover letter
- Mock interview questions from the job description
- Track applications on a Kanban board
On your resume
- Built an AI resume analyser (Next.js, Gemini API) that parses PDFs, scores them against job descriptions and suggests rewritten bullet points, with rate limiting and prompt-injection safeguards.