About the Project

Inspiration

Job hunting is often frustrating — endless scrolling, duplicate listings, and irrelevant results. Many talented people miss opportunities simply because they don’t have the time or tools to filter through thousands of postings.
We wanted to solve this by building an AI-powered job finder that acts like a personal career assistant: analyzing resumes, finding tailored jobs, and making applications effortless.

How We Built It

Backend Framework: Flask (Python)
Resume Analysis: Google’s Gemini AI to extract skills, roles, and experience level
Job Search Sources: LinkedIn, Indeed, Naukri.com, and JSearch API
Matching Algorithm:
- TF-IDF vectorization of resume/job descriptions
- Cosine similarity scoring:
  [ \text{similarity}(A, B) = \frac{A \cdot B}{|A| |B|} ]
- Experience-level bonus scoring for better personalization
Frontend: Modern dark-themed UI with glassmorphism, particle animations, and responsive design

What We Learned

How to integrate large language models (LLMs) for extracting structured insights from unstructured resumes
Techniques for semantic text matching using NLP (TF-IDF, cosine similarity)
Combining scraping + APIs to unify multiple job boards into one search engine
Importance of UI/UX in keeping users engaged — good design makes complex features feel simple

Challenges We Faced

Data Quality: Many job listings had incomplete or noisy descriptions; we had to clean and normalize them
Duplicate Jobs: The same listing often appeared across multiple platforms, so we built a deduplication pipeline
Rate Limits: APIs and scraping restrictions required careful request management and fallback strategies
Matching Accuracy: Tuning the similarity algorithm to give fair match scores without bias toward keyword stuffing
Deployment: Ensuring the app runs smoothly both locally and on cloud platforms with minimal setup