Projects
AI Powered Resume Maker
This project analyzes resumes with AI-assisted scoring to surface missing skills, improve structure, and deliver actionable recommendations. It integrates resume parsing, skill taxonomy matching, and NLP-based scoring to provide a clear improvement path. The focus was on building reliable data pipelines, clear feedback UX, and scalable API design for future model upgrades.
The problem this solves
Most job seekers don't know why their resume gets rejected. ATS systems silently filter them out for missing keywords, poor structure, or skill mismatches — with zero feedback. This project was built to change that by giving candidates an honest, AI-driven analysis of their resume against any job description.
What it does
- Resume parsing and structured data extraction
- Skill gap analysis against job description keywords
- NLP-based scoring across structure, clarity, and relevance
- Actionable improvement suggestions per section
- Gemini API integration for intelligent language feedback
- MongoDB persistence for storing analysis history
Hard parts and how I solved them
The hardest part was making the skill matching meaningful without being rigid. A resume that says 'built REST APIs with Node.js' shouldn't fail a check for 'Node.js experience with API design' — they're the same thing phrased differently. Solving this required semantic similarity rather than simple keyword matching, which made the Gemini integration non-trivial to get right. Getting the prompts to return structured, consistent JSON across wildly different resume formats also took significant iteration.
Tech stack
Outcome
The platform successfully parses resumes, scores them across multiple dimensions, and delivers specific, actionable feedback. Users can see exactly which skills are missing, which sections need improvement, and what a stronger version of their resume would look like — all within seconds of uploading.
My role
I built this end-to-end — from the resume parsing pipeline and Gemini API integration on the backend, to the scoring logic, the feedback display UI, and the MongoDB data layer. The product direction, API design, and UX decisions were all mine.