← all projects
2023 — 24ML + full-stackshipped

DeFraudAI

A misinformation-fighting platform: CNN-based deepfake detection served globally with low latency.

defraudai.app
PythonPythonTensorFlowTensorFlowFlaskFlaskReactReactCloudflare WorkersCloudflare Workers

01 / the problem

Deepfakes spread faster than fact-checkers can type. Detection needs to be accessible, fast, and global — not locked in a research notebook.

02 / what i built

  • Deepfake detection model built with TensorFlow transfer learning — 92% accuracy.
  • Flask API serving the model behind a React frontend for flagging and trust-score visualization.
  • Deployed with Cloudflare Workers for low-latency access worldwide.
  • Gemini-assisted text credibility analysis alongside the visual model.

Want the full walkthrough — architecture, decisions, war stories?

ask me about it