datamind.

AI-powered data analyst. Upload any file, get instant insights, charts, and answers — no analyst needed.

---

What is DataMind?

DataMind is a full-stack SaaS application that replaces the manual work of a data analyst with AI. You upload a file in any format — CSV, Excel, PDF, or an image — and within seconds you get a live PostgreSQL database, auto-generated visualizations, AI-written insights, and a natural language chatbot that converts plain English into SQL queries.

---

Architecture

┌─────────────────────────────────────────────────────────┐
│                     React Frontend                       │
│              (Vite + Tailwind + Recharts)                │
└────────────────────────┬────────────────────────────────┘
                         │ HTTP (REST API)
┌────────────────────────▼────────────────────────────────┐
│                   FastAPI Backend                        │
└──────┬─────────────────┬──────────────────┬─────────────┘
       │                 │                  │
┌──────▼──────┐  ┌───────▼──────┐  ┌───────▼──────┐
│   Layer 1   │  │   Layer 2    │  │   Layer 3    │
│  Ingestion  │  │  SQL Engine  │  │  Dashboard   │
│             │  │              │  │  + AI Insights│
│ CSV / Excel │  │  DataFrame → │  │              │
│ PDF / Image │  │  PostgreSQL  │  │  Gemini API  │
│ OCR / Parse │  │  (Supabase)  │  │              │
└─────────────┘  └──────────────┘  └──────────────┘
                         │
              ┌──────────▼──────────┐
              │  PostgreSQL Cloud   │
              │     (Supabase)      │
              │                     │
              │  • users            │
              │  • uploads          │
              │  • query_logs       │
              │  • dynamic tables   │
              └─────────────────────┘

---

Features

Layer 1 — Universal Data Ingestion

Accepts any file format and converts it into a clean, structured Pandas DataFrame. Handles messy real-world data including merged cells, multi-page PDFs, and low-resolution scanned images.

• CSV / Excel via Pandas
• PDF tables via pdfplumber
• Images via EasyOCR
• Automatic column name cleaning, type inference, and whitespace stripping

Layer 2 — Dynamic SQL Engine

Pushes any DataFrame into a live PostgreSQL database with automatic schema inference. Every upload creates a uniquely named, isolated table. All uploads are logged with UUIDs for full traceability.

• Dynamic table creation via SQLAlchemy
• Correct type inference (TEXT, BIGINT, FLOAT, TIMESTAMP)
• Upload tracking in a uploads metadata table
• Schema inspection for LLM context

Layer 3 — Auto Dashboard & AI Insights

Reads from the live PostgreSQL table and auto-generates a full visual dashboard — no hardcoding, works for any dataset. AI analysis is powered by Google Gemini.

• Auto bar charts, line charts, correlation heatmaps
• Key metric cards (sum, avg, min, max per numeric column)
• AI-written insights, trends, and business recommendations
• Raw data preview table

Layer 4 — English to SQL Chatbot (in progress)

A natural language interface where users type plain English questions and get SQL-powered answers back in real time.

• English → SQL conversion via LLM
• Runs queries against the live PostgreSQL table
• Returns results as tables or charts

---

Tech Stack

Layer	Technology
Frontend	React 18, Vite, Tailwind CSS, Recharts
Backend	FastAPI, Uvicorn
Database	PostgreSQL (Supabase)
ORM	SQLAlchemy
Data Processing	Pandas, NumPy
PDF Parsing	pdfplumber
OCR	EasyOCR
AI / LLM	Google Gemini API
Environment	Python 3.13, venv

---

Project Structure

datamind/
├── backend/
│   ├── main.py                  # FastAPI app entry point
│   └── routes/
│       ├── upload.py            # File upload endpoint
│       └── dashboard.py         # Dashboard data endpoint
├── frontend/
│   └── src/
│       ├── pages/
│       │   ├── Home.jsx         # Landing page
│       │   └── Dashboard.jsx    # Dashboard page
│       ├── components/
│       │   ├── MetricCard.jsx
│       │   ├── BarChart.jsx
│       │   ├── LineChart.jsx
│       │   ├── DataTable.jsx
│       │   └── InsightsCard.jsx
│       └── api/
│           └── client.js        # Axios API client
├── layers/
│   ├── layer1_ingestion/
│   │   └── ingestion.py         # File parsers
│   ├── layer2_sql/
│   │   └── sql_engine.py        # PostgreSQL engine
│   ├── layer3_dashboard/
│   │   └── insights.py          # Gemini AI insights
│   └── layer4_chatbot/          # Coming soon
├── utils/
│   └── database.py              # DB connection test
├── uploads/                     # Temporary file storage
├── .env                         # Environment variables (not committed)
└── requirements.txt

---

Getting Started

Prerequisites

• Python 3.10+
• Node.js 18+
• A Supabase account (free tier)
• A Google AI Studio API key (free tier)

1. Clone the repo

git clone https://github.com/DefinitelyMrityunjay/datamind.git
cd datamind

2. Set up the backend

python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

3. Configure environment variables

Create a .env file in the root:

DATABASE_URL=your_supabase_postgresql_connection_string
GEMINI_API_KEY=your_google_gemini_api_key

4. Set up the database

Run this SQL in your Supabase SQL editor:

CREATE TABLE IF NOT EXISTS users (
    user_id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
    email VARCHAR(255) UNIQUE NOT NULL,
    created_at TIMESTAMP DEFAULT NOW(),
    plan VARCHAR(50) DEFAULT 'free'
);

CREATE TABLE IF NOT EXISTS uploads (
    upload_id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
    user_id UUID REFERENCES users(user_id),
    file_name VARCHAR(255),
    file_type VARCHAR(50),
    table_name VARCHAR(255),
    uploaded_at TIMESTAMP DEFAULT NOW(),
    status VARCHAR(50) DEFAULT 'processing'
);

CREATE TABLE IF NOT EXISTS query_logs (
    query_id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
    user_id UUID REFERENCES users(user_id),
    upload_id UUID REFERENCES uploads(upload_id),
    english_input TEXT,
    generated_sql TEXT,
    executed_at TIMESTAMP DEFAULT NOW(),
    success BOOLEAN
);

5. Start the backend

cd backend
uvicorn main:app --reload

API runs at http://127.0.0.1:8000 API docs at http://127.0.0.1:8000/docs

6. Start the frontend

cd frontend
npm install
npm run dev

App runs at http://localhost:5173

---

Screenshots

---

Roadmap

• Layer 1 — Universal file ingestion (CSV, Excel, PDF, Image)
• Layer 2 — Dynamic PostgreSQL table generation
• Layer 3 — Auto dashboard with AI insights
• React frontend with FastAPI backend
• Layer 4 — English to SQL chatbot
• User authentication (Supabase Auth)
• Upload history page
• Cloud deployment (Vercel + Railway)
• Multi-sheet Excel support
• Export dashboard as PDF

---

Author

Mrityunjay — github.com/DefinitelyMrityunjay

---

Built as a portfolio project to demonstrate end-to-end data engineering, AI integration, and full-stack product development.