Data Flow

The data flow within Rakis follows a well-defined path, ensuring efficient processing and communication between components. Here is a high-level overview of the data flow:

Step 1: Inference Requests Inference requests originate from various sources, such as smart contracts or decentralized applications (dApps). These requests, containing prompts, security frames, and other parameters, are received by the P2P and Peering layer.

Step 2: Inference Management The Inference layer handles these requests by managing the scheduling, execution, and result processing through inference workers and embedding workers.

Step 3: Coordination The Coordination layer manages the commit-reveal process, timeout-based scheduling, and quorum building for the inference requests.

Step 4: Consensus Computation The Consensus layer computes the final consensus using embeddings and hashing algorithms, then propagates the consensus results to the network.

Step 5: Persistence The Persistence layer integrates with various blockchains, ensuring the persistence of inference results, implementing incentivization mechanisms, and supporting cross-chain operations.