About the project.
A zero-dependency vector database built from scratch in Java 17+ featuring side-by-side HNSW graph search, KD-Tree indexing, and Brute-Force search across Cosine, Euclidean, and Manhattan metrics. Integrated with a local RAG pipeline via Ollama (nomic-embed-text and llama3.2) for document ingestion, top-k context retrieval, and document-grounded answers. Includes a REST API for vector operations and search benchmarking alongside an interactive Web UI with live PCA vector projections.
Features
>HNSW, KD-Tree, and Brute-Force algorithms with Cosine, Euclidean, and Manhattan metrics.
Ollama integration with nomic-embed-text and llama3.2 for top-k context retrieval and document-grounded answers.
Interactive 2D vector visualization with live search benchmarking across graph and tree index structures.
Structured REST API endpoints delivering efficient CRUD operations, error handling, and validated data contracts.
Optimized database queries, clean architectural design, and minimal latency for production workloads.
Features
>Tech Stack
/>
JavaObject-oriented programming language for building robust, scalable enterprise backend applications.
OllamaFramework for running and embedding open-source large language models locally.
RAGRetrieval-Augmented Generation architecture combining vector search context with LLM inference.
Vector DatabasesSpecialized database engines engineered to store, index, and query high-dimensional vector embeddings.
Vector EmbeddingsHigh-dimensional numerical representations capturing semantic meaning of text and data.
REST APIsArchitectural style for designing structured, scalable networked endpoints using standard HTTP.
MavenBuild automation and dependency management tool primarily used for Java projects.
GitDistributed version control system for tracking changes and collaborating on codebases.
GitHubCloud-based hosting service for Git repositories with CI/CD workflows and collaboration.