01 / Project
Deepfake Detection using Blockchain
Web3 + AI fighting synthetic media, on-chain
deepfake-blockchain.stack
The Problem
Deepfakes are eroding digital trust. Centralized platforms can’t keep up with the volume of synthetic media or prove that a verdict hasn’t been tampered with.
What I Built
A Web3-enabled detection platform for a decentralized social media ecosystem. CNN and Vision Transformer models detect deepfakes, while detection results, perceptual hashes, and model embeddings are stored on IPFS and verified on Ethereum.
Approach
Frames are scored by an ensemble of CNN and ViT classifiers. Each verdict is hashed, pinned to IPFS, and anchored to an Ethereum smart contract so any third party can independently audit a detection result without trusting a central server.
Next project
Production RAG Pipeline (Dvara ML)