TransformDash

Hybrid Data Transformation & Dashboard Platform

Run SQL transformations with dependency management and lineage tracking directly against PostgreSQL, without needing a data warehouse.

---

Try the Live Demo

Live Demo

Login options:

• Viewer: demo / demo (read-only on source data, can run transformations)
• Admin: admin / admin (full access)

Experience TransformDash without installing anything:

• Pre-loaded with 100 customers, 500 orders across 24 tables
• Run transformations and see Bronze → Silver → Gold pipeline in action
• Build interactive dashboards and explore data visualizations
• Test ML model predictions on sample data

Demo Notes:

• Shared demo environment - source data is protected (read-only)
• You can run transformations and create dashboards
• First load takes 30-60 seconds if the server is sleeping (free tier)
• Demo data maintained automatically

---

Features

Core Capabilities

• Multi-Layer Architecture: Bronze → Silver → Gold medallion pattern
• SQL & Python Models: SQL with Jinja templating and Python transformations
• DAG Orchestration: Automatic dependency resolution and parallel execution
• Interactive Web UI: Real-time lineage graphs and dashboards
• PostgreSQL Support: Full support for transformations
• Incremental Syntax: Write incremental models (full refresh for now, true incremental on roadmap)

AI-Powered Search (Optional)

• Semantic Search: Natural language queries to find models (e.g., "customer revenue models")
• FAISS Vector Search: Fast similarity search using sentence embeddings
• Smart Model Discovery: Search by meaning, not just keywords
• Graceful Degradation: Optional feature - install dependencies only if needed
• Installation: pip install -r dbt_assistant/requirements.txt
• See dbt_assistant/README.md for details

Model Features

• {{ source() }} and {{ ref() }} macros
• {{ config() }} for model configuration
• {% if is_incremental() %} syntax support (currently does full refreshes)
• YAML-based source definitions
• View and table materializations

---

Architecture

┌─────────────────────────────────────────────────────────────┐
│                      TransformDash                          │
├─────────────────────────────────────────────────────────────┤
│                                                             │
│  Raw Sources (PostgreSQL)                                   │
│         ↓                                                   │
│  Bronze Layer (stg_* models - Views)                       │
│    • Direct extraction from raw tables                      │
│    • Column aliasing and standardization                    │
│         ↓                                                   │
│  Silver Layer (int_* models - Tables)                      │
│    • Multi-table joins                                      │
│    • Business logic and calculations                        │
│    • Aggregations and window functions                      │
│         ↓                                                   │
│  Gold Layer (fct_*/dim_* models - Tables)                  │
│    • Analytics-ready fact and dimension tables              │
│    • Final business metrics                                 │
│         ↓                                                   │
│  Web Dashboard & API                                        │
│    • Interactive lineage visualization                      │
│    • Model catalog and documentation                        │
│                                                             │
└─────────────────────────────────────────────────────────────┘

---

Quick Start

Fastest Way to Try It (Docker - Recommended)

# Clone the repository
git clone https://github.com/kraftaa/transformdash.git
cd transformdash

# Setup environment for Docker
cp .env.docker .env
# Generate JWT secret and add to .env
python -c 'import secrets; print("JWT_SECRET_KEY=" + secrets.token_urlsafe(32))' >> .env

# Start all services (PostgreSQL + TransformDash)
docker-compose up -d

# Wait for containers to start (about 10 seconds)
sleep 10

# Run database migrations
docker-compose exec web bash run_migrations.sh

# Load sample data
docker-compose exec web python load_sample_data.py

# Train example ML model
docker-compose exec web python ml/train_telco_churn.py

Then visit http://localhost:8000 (default login: admin / admin)

What you get:

• PostgreSQL database with user authentication
• Sample e-commerce dataset (24 tables, 100+ customers, 500+ orders)
• Trained ML model (Telco Customer Churn with realistic metrics)
• Interactive dashboards and chart builder
• ML Models tab with prediction capabilities

Install from PyPI (Simplest)

# Install the package
pip install transformdash

# Set up PostgreSQL (required - choose one option):

# Option A: Use Docker for PostgreSQL
docker run -d --name transformdash-db \
  -p 5432:5432 \
  -e POSTGRES_PASSWORD=mypassword \
  -e POSTGRES_DB=transformdash \
  postgres:15

# Option B: Use your existing PostgreSQL server
# (Make sure you have PostgreSQL 15+ running)

# Create .env file with your database credentials
cat > .env << 'EOF'
# Generate this: python -c 'import secrets; print(secrets.token_urlsafe(32))'
JWT_SECRET_KEY=your-secret-key-here

# Main TransformDash database (stores dashboards, charts, users)
TRANSFORMDASH_HOST=localhost
TRANSFORMDASH_PORT=5432
TRANSFORMDASH_DB=transformdash
TRANSFORMDASH_USER=postgres
TRANSFORMDASH_PASSWORD=mypassword

# Your analytics database (the data you want to analyze)
APP_HOST=localhost
APP_PORT=5432
APP_DB=production
APP_USER=postgres
APP_PASSWORD=mypassword
EOF

# Start the application
python -m ui.app

# Visit http://localhost:8000 (login: admin / admin)

Important Notes:

• PostgreSQL is required - TransformDash is a PostgreSQL-based platform
• CSV upload is available through the UI, but files are loaded into PostgreSQL tables
• You cannot use TransformDash without a PostgreSQL database
• The app provides dbt-like transformations + interactive dashboards for PostgreSQL

Local Development Setup (Without Docker)

# Clone the repository
git clone https://github.com/kraftaa/transformdash.git
cd transformdash

# Create virtual environment
python3 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Start PostgreSQL with Docker (just the database)
docker-compose up -d db

# Wait for database to start
sleep 5

# Run migrations
bash run_migrations.sh

# Load sample data
python load_sample_data.py

# Train ML model
PYTHONPATH=. python ml/train_telco_churn.py

# Start the application
python ui/app.py  # Visit http://localhost:8000

# Login: admin / admin

Prerequisites

• Docker & Docker Compose (for Quick Start)
• OR Python 3.9+ and PostgreSQL 15+ (for manual installation)
• Git

Installation Options

Option 1: Docker Compose (Recommended)

# Clone the repository
git clone https://github.com/kraftaa/transformdash.git
cd transformdash

# Generate a secure JWT secret key
python -c 'import secrets; print(secrets.token_urlsafe(32))' > jwt_key.txt
export JWT_SECRET_KEY=$(cat jwt_key.txt)

# Start all services (includes PostgreSQL)
docker-compose up -d

# Access at http://localhost:8000

Option 2: Install via pip (From Source)

# Clone the repository
git clone https://github.com/kraftaa/transformdash.git
cd transformdash

# Create virtual environment
python3 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install all dependencies (including ML support)
pip install -r requirements.txt

# Set up environment variables
cp .env.example .env
# Generate a secure JWT secret key
python -c 'import secrets; print(secrets.token_urlsafe(32))'
# Add the key to .env as: JWT_SECRET_KEY=<generated-key>

# Run the web UI
python ui/app.py

Option 3: Docker Only

# Build and run with your own PostgreSQL
docker build -t transformdash:latest .
docker run -d -p 8000:8000 \
  -e TRANSFORMDASH_HOST=your-postgres-host \
  -e TRANSFORMDASH_PASSWORD=your-password \
  transformdash:latest

Option 4: Kubernetes (Production)

# See DEPLOYMENT.md for full instructions
kubectl apply -f k8s/

Option 5: Testing Kubernetes Configs Locally (Minikube)

# Start minikube
minikube start

# Build Docker image in minikube's Docker environment
eval $(minikube docker-env)
docker build -t transformdash:latest .

# Deploy to minikube
kubectl apply -f k8s/

# Access the application
minikube service transformdash-service -n transformdash

For detailed deployment instructions, see DEPLOYMENT.md

Configuration

1. Set up database credentials:

cp .env.example .env
# Edit .env with your database credentials

Example .env:

TRANSFORMDASH_HOST=localhost
TRANSFORMDASH_PORT=5432
TRANSFORMDASH_DB=transformdash
TRANSFORMDASH_USER=postgres
TRANSFORMDASH_PASSWORD=your_password

APP_HOST=localhost
APP_PORT=5432
APP_DB=production
APP_USER=postgres
APP_PASSWORD=your_password

2. Initialize databases (if not using Docker Compose):

createdb transformdash
createdb production

Run Your First Transformation

Method 1: Web UI (Recommended)

# Start the web interface
python ui/app_refactored.py
# Visit http://localhost:8000

# Navigate to Models and click "▶️ Run Models"

Method 2: Create Charts and Dashboards

# Access the UI at http://localhost:8000
# 1. Go to "Chart Builder" to create visualizations
# 2. Go to "Dashboards" to build interactive dashboards
# 3. Use filters and drill-downs for analysis

Method 3: Train ML Models

# Train an example model
PYTHONPATH=. python ml/examples/train_example_model.py

# View registered models
PYTHONPATH=. python ml/registry/model_registry.py

# Use models in SQL transformations (see ml/README.md)

Method 4: API Access

# View API documentation
open http://localhost:8000/docs

# Execute transformations via API
curl -X POST http://localhost:8000/api/models/execute

# Query data
curl -X POST http://localhost:8000/api/query \
  -H "Content-Type: application/json" \
  -d '{"table": "my_model", "limit": 100}'

---

📁 Project Structure

transformdash/
├── connectors/              # Database connectors
│   ├── redis.py            # Redis connector
│   └── (mongodb, etc.)
├── dbt_assistant/          # Optional AI search module
│   ├── core.py            # AI search assistant
│   ├── parser.py          # SQL model parser
│   ├── embed_search.py    # FAISS semantic search
│   ├── requirements.txt   # Optional dependencies
│   └── README.md          # AI search documentation
├── models/                  # SQL transformation models
│   ├── sources.yml         # Data source definitions
│   ├── bronze/             # Staging layer (stg_*)
│   │   ├── stg_customers.sql
│   │   └── stg_orders.sql
│   ├── silver/             # Intermediate layer (int_*)
│   │   └── int_customer_orders.sql
│   └── gold/               # Analytics layer (fct_*, dim_*)
│       └── fct_orders.sql
├── transformations/         # Core transformation engine
│   ├── model.py            # Transformation model class
│   ├── dag.py              # DAG builder and validator
│   └── model_loader.py     # SQL model loader
├── orchestration/           # Execution engine
│   └── engine.py           # DAG orchestrator
├── ui/                      # Web interface
│   └── app.py              # FastAPI application
├── tests/                   # Test suite
├── config.py               # Environment configuration
├── postgres.py             # PostgreSQL connector
├── requirements.txt         # Python dependencies
└── README.md               # This file

---

Creating Models

Bronze Layer (Staging)

File: models/bronze/stg_customers.sql

{{ config(materialized='view') }}

-- Bronze layer: Direct extraction with minimal transformation

with transformed_data as (
    select
        id as customer_id,
        email,
        name as customer_name,
        created_at
    from {{ source('raw', 'customers') }}
)

select * from transformed_data

Silver Layer (Intermediate)

File: models/silver/int_customer_orders.sql

{{ config(
    materialized='incremental',
    unique_key='order_id'
) }}

-- Silver layer: Join customers with orders

with transformed_data as (
    select
        o.order_id,
        o.customer_id,
        c.customer_name,
        c.email as customer_email,
        o.order_date,
        o.total_amount
    from {{ ref('stg_orders') }} o
    join {{ ref('stg_customers') }} c
        on o.customer_id = c.customer_id

    {% if is_incremental() %}
        -- Only process new orders
        where o.order_date > (select max(order_date) from {{ this }})
    {% endif %}
)

select * from transformed_data

Gold Layer (Analytics)

File: models/gold/fct_orders.sql

{{ config(materialized='table') }}

-- Gold layer: Final fact table

with transformed_data as (
    select
        order_id,
        customer_id,
        customer_name,
        order_date,
        total_amount,
        extract(year from order_date) as order_year,
        extract(month from order_date) as order_month
    from {{ ref('int_customer_orders') }}
)

select * from transformed_data

---

Web UI Features

Dashboard

• Model Catalog: Browse all transformation models
• Layer Statistics: Bronze/Silver/Gold model counts
• Real-time Updates: Refresh models dynamically
• Code Viewer: Click any model to see its SQL code
• Run Transformations: One-click execution of entire DAG

Lineage Graph

• Interactive Visualization: D3.js-powered lineage graphs
• Dependency Tracking: See how models depend on each other
• Color-Coded Layers: Bronze (🟫), Silver (⚪), Gold (🟡)

Execution

• ▶️ Run Button: Execute all transformations in DAG order
• Status Tracking: See execution progress and results
• Error Handling: Clear error messages if something fails
• Metrics: Total time, successes, failures

API Endpoints

• GET /: Interactive dashboard
• GET /api/models: List all models with dependencies
• GET /api/models/{name}/code: Get SQL code for a model
• POST /api/execute: Run all transformations
• GET /api/lineage: Get DAG structure
• GET /api/health: Health check

---

Configuration

Environment Variables (.env)

# PostgreSQL
POSTGRES_HOST=localhost
POSTGRES_PORT=5432
POSTGRES_DB=your_database
POSTGRES_USER=your_user
POSTGRES_PASSWORD=your_password

# MongoDB (optional)
MONGO_URI=mongodb://localhost:27017
MONGO_DB=your_mongo_db

# Redis (optional)
REDIS_HOST=localhost
REDIS_PORT=6379
REDIS_DB=0

Sources Configuration (models/sources.yml)

version: 2

sources:
  - name: raw
    description: "Your raw data source"
    database: your_database
    schema: public
    tables:
      - name: customers
        columns:
          - name: id
            tests:
              - not_null
              - unique
          - name: email

---

Testing

# Run unit tests
pytest tests/

# Test database connection
python postgres.py

# Test model loader
python transformations/model_loader.py

# Run example pipeline
python run_transformations.py

---

Development

Adding a New Database Connector

1. Create connector class in connectors/:

class MyDatabaseConnector:
    def __init__(self, connection_string):
        self.conn = ...

    def query_to_dataframe(self, query):
        return pd.read_sql(query, self.conn)

2. Add to config.py:

MY_DB_URI = os.getenv('MY_DB_URI')

3. Use in transformations:

from connectors.mydatabase import MyDatabaseConnector

def my_transformation(context):
    with MyDatabaseConnector() as db:
        return db.query_to_dataframe("SELECT * FROM table")

Adding Custom Macros

Extend ModelLoader in transformations/model_loader.py:

def my_custom_macro(self, arg1, arg2):
    return f"processed_{arg1}_{arg2}"

# Register in render_sql method
env.globals['my_macro'] = self.my_custom_macro

---

Use Cases

Data Warehousing

• Extract data from multiple sources
• Transform with SQL for performance
• Load into analytics-ready tables

Business Intelligence

• Create conformed dimensions
• Build fact tables for metrics
• Serve dashboards and reports

Data Engineering

• Orchestrate complex pipelines
• Track data lineage
• Incremental processing for efficiency

Analytics Engineering

• SQL transformations with dependency management
• Version-controlled SQL
• Collaborative data modeling

---

Contributing

Contributions are welcome! Please follow these steps:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

---

License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Acknowledgments

• Built after working with dbt, Airflow, Airbyte, Superset, Tableau, and building/running custom Rust transformations in Kubernetes via CronJobs - wanted a single tool that combines transformation, orchestration, and visualization
• Built with FastAPI, Pandas, and D3.js
• Follows the Medallion Architecture pattern

---

Support

• Documentation: See GETTING_STARTED.md and DEPLOYMENT.md
• Issues: GitHub Issues
• Discussions: GitHub Discussions

---

Roadmap

• Building transformdash pip package
• Add Spark connector for big data
• Add Netsuite connector
• Add S3 connector
• Implement data quality testing framework
• Add CI/CD pipeline templates
• Create VSCode extension
• Real-time data streaming
• Cloud deployment guides (AWS, GCP, Azure)
• Metric computation layer
• Row-level security

---