How deven96/ahnlich works
Main Components Identification
a) Core Services:
• Ahnlich DB (Vector Store Service)
• Ahnlich AI (AI Proxy Service)
• Client SDKs (Python and Rust)
• CLI Interface
b) Supporting Components:
• gRPC Communication Layer
• Persistence Layer
• Model Cache Storage
• Tracing System (Jaeger)
Architectural Layers
a) Client Layer:
• Both Python and Rust SDK clients
• CLI interface
• Web interface (from web/ahnlich-web)
b) Service Layer:
• AI Service (port 1370)
• DB Service (port 1369)
c) Processing Layer:
• Vector Processing
• AI Model Processing
• Similarity Search Algorithms
d) Storage Layer:
• In-memory Vector Store
• Persistence Storage
• Model Cache
Component Relationships
• Client-to-Service communication via gRPC
• AI Service to DB Service communication
• Storage interactions
• Tracing flow
Specific Implementation Details
Technical details such as:
• Supported similarity algorithms (cosine, euclidean, kdtree)
• AI model support (all-minilm-l6-v2, resnet-50)
• Protocol specifications (gRPC)
• Persistence mechanisms
Additional Elements
Include:
• Port numbers
• Supported execution providers (CUDA, CoreML)
• Configuration options
• Persistence paths
• Model cache locations
Notes
Important notes regarding:
• Production readiness status
• Required dependencies (CUDA v12, etc.)
• Persistence configuration
• Tracing setup
Remember to indicate the scalability aspects and how the system handles different types of queries (vector similarity search, metadata filtering, AI model processing) through the various components.
This design should reflect both the current implementation and allow for future extensions as indicated by the project's structure and documentation.