FakeOut-AI

FakeOut uses Gemini AI to detect fake job listings and warn users before they get scammed. It analyzes job posts, flags red signals, and provides a confidence score — all in seconds.

Comment

🧠 Inspiration

A few months back, I came across someone in my circle who fell victim to a fake job listing. They paid a registration fee and got ghosted. Since then, I started noticing how common these scammy job posts are — especially on WhatsApp groups, Telegram, and sketchy job boards. They often promise high pay, no experience, and ask for upfront fees. I felt like... someone should build a simple tool to catch this. So I decided — why not me?

That's how FakeOut started — not just as a hackathon idea, but as a personal itch to build something useful, something that could actually help people.

💡 What it does

FakeOut is an AI-powered tool that helps people detect fake or suspicious job listings. You just paste the job description (or upload a .txt file), and it analyzes it using Google’s Gemini AI. It then:

  • Gives a confidence score
  • Highlights any red flags
  • Tells you whether to Apply, Avoid, or Proceed with caution

All this happens in a clean, minimal UI — accessible to anyone, even those who aren’t tech-savvy.

🛠️ How we built it

I built FakeOut solo using:

  • Python + Streamlit for the frontend
  • Google Gemini (via Generative AI API) for the brain behind it
  • UV (ultra fast Python package manager) to set everything up quickly
  • Hosted the final project on Streamlit Cloud

The logic is prompt-engineered carefully to make Gemini act cautiously, call out scam triggers, and even explain why a job post might be suspicious.

⚔️ Challenges we ran into

So many. I’ll be honest — it wasn’t always smooth:

  • False positives and false negatives from the AI — at one point it marked a clearly fake job as “genuine with 1.0 confidence”! That forced me to rethink the prompt and scrap some rule-based scoring.
  • Balancing LLM output and user trust — you can’t just say “AI said so”, you need to communicate risk smartly.
  • Low specs laptop — no GPU, no fancy IDEs. I had to make everything work with minimal setup, lightweight tools, and zero bloat.
  • Getting the UX just right so it doesn’t feel like a tech demo, but an actual product someone would use.

But I pushed through. Because I believed in the idea.

🏁 Accomplishments that we're proud of

  • I built this end-to-end, solo, in just a few days.
  • The AI response is accurate, well-explained, and actually useful.
  • It’s already tested with real fake jobs — and it caught them!
  • The entire app is deployed, polished, and ready to use.
  • Honestly? Just seeing it go from an idea in my head to a working tool… that’s a win.

📚 What we learned

This project pushed me to:

  • Think like a product builder, not just a coder.
  • Work with AI prompt design seriously — which is honestly an art.
  • Understand how to communicate risk to users in a way that feels helpful, not scary.
  • Polish every detail — from config files to captions — because it all matters when you're building solo.

It taught me that even a single dev, with limited resources, can ship something meaningful.

🚀 What's next for FakeOut-AI

  • Add a browser extension to scan job listings on websites instantly
  • Train a fine-tuned model on thousands of scam vs real job listings for offline use
  • Let users report scam jobs and crowdsource intelligence
  • And maybe one day... turn it into a real product that saves people from losing money and trust

But for now — I’m proud of what I’ve built. And excited to show it here.

Thanks for reading. 🙏

Built With

Share this project:

Updates