Powered by blockchain integrated
federated / machine learning
The Terracuda.Ai Platform
Our platform support industries, to utilize data driven digital transformation by establishing trust to collaborate, share and transact securely. Every company is a data driven company, that is central to their business strategy. Within organizations and cross organizations, data remains “silo”, lacking the trust to share. Governments around the world have strict regulations in place around data privacy restricts sharing. When data are applied, there are questions around data validity and integrity for strategic decision making. The Terracuda.Ai platform, combining private permissioned blockchain with federated learning and AI/ML orchestration is here to solve this dilemma.
This platform capability can be used for collaborative development of highly accurate machine learning models. Through federated learning, the models can be trained locally on distributed datasets without the need to move any sensitive data. The guarantees the privacy preservation of the data, the model itself and the out put of the model, allowing for secure data exchange. Blockchain on the other hand, ensures the data security, traceability and integrity of these models. This integrated solution will empower enterprise competitiveness, discover innovate business models, and automate services with efficiency, within the regulatory data framework.
Our platform integrated features
Permissioned blockchain
Data security, traceability and integrity within business units or cross organizations
Transactions automated via smart contracts
Consortium management access
AI / ML orchestration
Simply execute algorithms on data stored in multiple locations without actual data transfer, using federated learning
Supports centralized / edge cloud
External AI / ML through API
Our management & orchestration solution
(global model)
Model distribution
Aggregate weight
Blockchain traceability
Local storage
Local storage
Enterprise 1
Enterprise 2
Our Terracuda.Ai solutions for life science
Key highlights of Terracuda.Ai
All operations on the platform are written onto an immutable decentralized ledger that contains only non-sensitive metadata. This include :
Anonymous identifiers of assets on the network, including datasets
Associated permission for using assets, including datasets
Specification of training, aggregation and evaluation tasks, constituting compute plans. Ensuring both traceability and reproducibility
Data is never shared on the Terracuda.Ai platform, only the algorithm & the model weight. We’re enabling enterprise privacy by integrating the following technologies:
Traceability of access to algorithms and models
Zero-Knowledge Proof (ZKP) cryptography(separate privacy layer) integrated solution is an added powerful tool to the platform, that can be tailored for deployment across various use cases and requirements
Secure Multiparty Computation (SMC) cryptographic protocol that distributes a computation across multiple parties. Enabling scientists, analysts to compliantly securely and privately compute on distributed data without ever exposing or moving it. (Experimental)
Security is at the heart of our solutions. We have designed enterprise partnership permission & roles in guaranteeing the security of the platform with the highest level of privacy.
We enforce Identity and access management (IAM) capabilities for user on-boarding
Certificate Authority (CA) provides secure services related to user enrollment, transactions invoked on the blockchain (guarantees the authenticity of the public key) and TLS secured connections
Leveraging Trusted Platform Modules (TPM) for sensitive code execution. Applying trusted environments through our HSM (Hardware Security Module) which is a dedicated cryptographic processor to protect highly critical and sensitive keys and assets with controlled access for members. We offer various applied hardware options
Enforce access control in smart contracts with applied endorsement business policies
Data is encrypted at rest
All network encryptions are encrypted applying standard encryption procedures and libraries. For communication across nodes, we have (A) secure GRPC protocol for non sensitive metadata exchange between orchestrators, encrypted with TRS (B) HTTPS Rest API for communication between node backend (models and models updates)
Model updates can be encrypted to ensure that given a centralized aggregator node in certain circumstances, cannot access sensitive data.