Here is a simplified flow.
- Train the model
import torch
model = torch.nn.Linear(2, 1) # simple linear model
2. Compile the model in a proven circuit
Using a Giza or other compiler, converts the model to Cairo (or ZK -friendly format).
giza compile model.onnx --output model.cairo
3. Create execution reasoning + proof
giza prove --input input.json --model model.cairo --output proof.json
4. Confirmation of evidence (Onchain)
@external
fn verify_prediction(proof: Proof) -> bool
assert(is_valid_proof(proof));
return true;
Health care: Personal diagnosis verification
financial resources: Validable credit score
gambling: AI logic for enemies without deception
Identity: Personal or repeated proof
Oracle: Onchain Defi platform using ML signal
This is not approval. This is part of what I explored while learning.
Journalist
- Compile the pytorch model with Cairo
- Check the ML output in Starknet
ezkl
- The onnx model uses Halo2 -based proof
Modular Lab
- Research -oriented ZKML tooling