Anansi
Distributed Computing
Anansi for GPU ML Training enables distributed machine learning across trusted execution environments. Anansi Distributed provides multi-party computation where no single party can cheat, using clean rooms and TEEs to prove computational integrity for healthcare companies and other organizations requiring secure collaboration.
Runs on AWS • Azure • GCP • Databricks • Snowflake
Secure Collaboration Through Trusted Execution
Multi-party computation with cryptographic proof generation enables secure collaboration across organizations. Healthcare companies can collaborate on research without exposing sensitive data, using TEEs to prove computational integrity and maintain regulatory compliance.
Multi-party Trust
Healthcare institutions can collaborate on ML models without sharing patient data. TEEs prove no single party can cheat or access others' data.
Clean Room Computing
Trusted execution environments create verifiable clean rooms where sensitive computations happen with cryptographic proof of integrity.
GPU ML Training
Distributed machine learning across multiple GPUs with hardware-level attestation. Train models collaboratively while maintaining data sovereignty.
For Engineers
Three steps. Takes about two minutes to set up. Works with whatever you're already doing.
Install Anansi
Works with your existing setup
Just a pip install. No config files, no setup scripts, no headaches.
pip install anansi-computeWrap Your Function
One line change
Your code stays the same. Just wrap it and get cryptographic proof of execution.
import anansi
result = anansi.compute(your_function, data, proof=True)Get Your Proof
Tamper-evident results
Standard cryptographic proof that anyone can verify. No trust required.
{
"result": your_computed_result,
"proof": "0x7f8a9b2c3d4e5f6a...",
"attestation": "verified",
"timestamp": "2025-01-20T23:33:38Z"
}For Compliance
Every computation generates a cryptographic proof bundle. Here's what auditors see.
Proof Bundle Structure
Job Identity
Who ran it, what code, which data inputs
Execution Environment
Cloud, region, instance type, timing
Cryptographic Attestation
CPU and GPU hardware verification
Digital Signature
Tamper-evident seal over the entire bundle
Sample Proof Bundle
{
"version": "1.0",
"proof_id": "prf_2025-08-21_7c2a",
"job": {
"job_id": "dbx-13482",
"caller": "analyst@company.com",
"entrypoint": "s3://bucket/ml-training:sha256:3f8f...e21"
},
"route": {
"cloud": "azure",
"region": "eastus",
"instance": "NCCads_H100_v5"
},
"timing": {
"started_at": "2025-08-21T01:03:12Z",
"ended_at": "2025-08-21T01:07:55Z"
},
"attestation": {
"cpu": {
"type": "SEV-SNP",
"verified": true
},
"gpu": {
"type": "NRAS",
"verified": true
}
},
"signature": {
"alg": "EdDSA",
"verified": true
}
}