Welcome to flora64, a revolutionary database that brings SQL-like simplicity to multimodal data handling. No more complex pipelines - just write familiar operations, and let flora64 handle the scalability and optimization behind the scenes.

What is flora64?

flora64 is a versatile database system that turns complex multimodal operations into simple queries:

• Handle images, text, video, and audio as easily as numbers in SQL
• Write familiar operations like GROUP BY and AVG for multimodal data
• Let the database automatically optimize vector operations
• Focus on your application logic, not infrastructure

Key Features

🚀 Multimodal Support

Built from the ground up to handle diverse data types:

• Text processing with built-in operations
• Image storage and retrieval
• Video and audio data management
• Vector embeddings support

💡 SQL-Like Simplicity

• Query multimodal data like traditional databases
• Built-in pipeline management and optimization
• Familiar operations: GROUP BY, AVG, JOIN for vectors
• No infrastructure complexity to manage

⚡ Automatic Pipeline Management

• Zero pipeline setup required
• Automatic scaling and optimization
• Built-in data transformation handling
• Smart resource management

Quick Start

from flora64 import Db, Text, Image

# Create a database instance with automatic pipeline management
db = Db()

# Define a multimodal table - as simple as creating a regular SQL table
posts = db.create_table({
    "title": Text,
    "content": Text,
    "image": Image,
    "category": Text
})

# Insert data - flora64 handles all the pipeline complexity
posts.insert([
    {
        "title": "Mountain Adventure",
        "content": "A beautiful sunset over the mountains",
        "image": "sunset.jpg",
        "category": "nature"
    },
    {
        "title": "City Life",
        "content": "Urban architecture at night",
        "image": "city.jpg",
        "category": "urban"
    }
])

# SQL-like operations on multimodal data
results = posts.query({
    # GROUP BY with vector aggregations - just like SQL!
    "category_vectors": posts.group_by("category").agg({
        "avg_embedding": posts["content"].embeddings().mean(),
        "image_cluster": posts["image"].embeddings().cluster_center()
    }),
    
    # JOIN with vector similarity - as simple as a regular JOIN
    "similar_posts": posts.join(
        posts,
        on=lambda x, y: x["content"].embeddings().cosine_similarity(
            y["content"].embeddings()
        ) > 0.8
    ),
    
    # WHERE clause with semantic search
    "nature_posts": posts.where(
        posts["content"].embeddings().similar_to("mountain landscape")
    ),
    
    # Aggregations across modalities
    "cross_modal_score": posts.group_by("category").agg({
        "text_image_alignment": (
            posts["content"].embeddings()
            .cosine_similarity(posts["image"].embeddings())
            .mean()
        )
    })
})

# All the complex pipeline operations are handled automatically:
# - Model loading and optimization
# - Batch processing for efficiency
# - Memory management
# - GPU acceleration when available
# - Caching and indexing
# - Pipeline parallelization

Advanced Features

Vector Operations

• Generate and store embeddings
• Cosine similarity calculations
• Vector math operations (dot product, normalization)
• Efficient batch processing

SQL-Style Operations for Multimodal Data

• GROUP BY on image features or text embeddings
• AVG/SUM operations on vector spaces
• JOIN operations across different modalities
• Aggregate functions for embeddings

Automated Pipeline Management

• Zero-config scaling for large datasets
• Automatic optimization of vector operations
• Built-in caching and indexing
• Smart resource allocation for heavy computations

Use Cases

1. AI/ML Applications

   • Store and process training data
   • Manage embeddings for semantic search
   • Handle multimodal inputs for ML models

2. Content Management

   • Process and store rich media content
   • Text analysis and processing
   • Image and video management

3. Data Processing Pipelines

   • ETL operations
   • Real-time data transformations
   • Batch processing

Getting Started

Ready to start using flora64? Here’s how:

1. Visit our GitHub repository to get the latest release

We’re excited to see what you’ll build with flora64!