CAREERSCOUT-AI

Authors

  • 1A Satyanarayana, 2 V Rajesh, 3 S Amaresh, 4 V Uday Kumar,5 R Aravind Author

DOI:

https://doi.org/10.62643/

Abstract

CareerScout-AI is an advanced AI-powered internship and hackathon discovery system that automatically scrapes multiple job portals, filters relevant opportunities using a Large Language Model (LLM), and delivers personalized results directly to users via a Telegram Bot — completely free and fully automated. With the rapid growth of artificial intelligence and automation technologies, there is a growing need for intelligent tools that save students valuable time in finding career opportunities. The system uses Playwright and BeautifulSoup to scrape six major platforms — Internshala, Unstop, Devfolio, MLH, LinkedIn, and Google Careers — collecting 50+ listings daily. A Groq-powered LLM (Llama-3.1-8b-instant) then runs a 3-step prompt chain: structured data extraction, relevance scoring (0-10), and personalized Telegram message generation. Only listings scoring 7 or above are recommended and broadcast to subscribers. The system is fully automated using GitHub Actions, which triggers the pipeline every day at 9:30 AM IST. A FastAPI backend exposes REST endpoints for manual pipeline runs, resume ATS analysis, and Telegram webhook handling. A Supabase PostgreSQL database stores all filtered jobs and subscriber preferences. Users can set domain filters (AI/ML, Web Dev, Data Science, Cybersecurity, Mobile Dev) and receive only relevant listings. Additionally, the system includes an ATS Resume Optimizer that accepts a PDF resume, extracts text using PyMuPDF, and uses Groq AI to return an ATS match score (0-100), matching and missing keywords, improvement suggestions, and an auto-generated cover letter. Overall, CareerScoutAI demonstrates the practical application of GenAI in automating career discovery for undergraduate students.

Downloads

Published

12-06-2026

How to Cite

CAREERSCOUT-AI. (2026). International Journal of Engineering Research and Science & Technology, 22(2(1), 2621-2629. https://doi.org/10.62643/