🌱 GreenOps: AI-Driven ESG & Carbon Accounting Engine

GreenOps is an enterprise-grade Environmental, Social, and Governance (ESG) compliance and predictive analytics engine. Engineered specifically for SMEs (Small and Medium-sized Enterprises), it digitizes the GHG Protocol to track Scope 1, 2, and 3 emissions while utilizing autonomous AI agents to ensure international regulatory compliance (e.g., EU CBAM).

🚀 Core Capabilities

📊 GHG Protocol Physics Engine

• Strict Regional Math: Calculates carbon footprints using geofenced, IPCC/DEFRA-compliant emission factors.
• Scope Isolation: Automatically categorizes entries into Scope 1 (Direct), Scope 2 (Purchased Electricity), and Scope 3 (Value Chain).
• Custom PPAs: Supports enterprise Power Purchase Agreements (PPAs) with self-verified custom grid overrides.

🧠 Autonomous AI Advisory (Llama-3)

• Context Compression: Dynamically compresses massive Pandas DataFrames to bypass LLM token limits without losing mathematical accuracy.
• Regulation Radar: Cross-references facility locations and export markets against emerging carbon border taxes (e.g., EU CBAM, US frameworks).
• Optimization Strategy: Generates real-time CapEx/OpEx operational optimization and verified offset recommendations.

🛡️ Enterprise Data Integrity

• Cryptographic Ingestion: Uses MD5 row-hashing during bulk CSV uploads to detect and block duplicate data corruption.
• Zero-Ghost Data: Server-side validation actively blocks 0.0 quantity logging to prevent artificial deflation of annual run-rate projections.
• Surgical Range Delete: Destructive database actions are protected by explicit user verification protocols.

📑 Executive Reporting

• Memory-Safe Architecture: Bypasses raw-row rendering limits by synthesizing data via FPDF into high-level, 1-page executive PDF summaries.
• Ternary Analytics: Maps energy transition trajectories (Clean vs. Grid vs. Fossil) using complex Plotly ternary scatter charts.

---

🏗️ System Architecture

GreenOps utilizes Streamlit's native multi-page architecture to isolate memory and ensure zero-latency routing.

GreenOps/
├── Home.py                     # UI Landing Page & SaaS Hero Section
├── data_store.py               # Session State Management & Data Persistence
├── ui_components.py            # Reusable UI Components & CSS Styling
├── emission_factors.py         # The Physics Engine: Centralized IPCC/DEFRA constants
├── ai_agents.py                # LLM Interface: CrewAI logic via Groq
├── report_generator.py         # PDF Report Generation Engine (FPDF2)
├── data_generator.py           # Synthetic Data Generation Utilities
├── pyproject.toml              # Project Dependencies & Configuration (uv)
├── .streamlit/
│   └── config.toml             # Theme Config: Hardcoded Light Mode to prevent OS clashing
├── fonts/                      # UTF-8 Font binaries for FPDF export (DejaVu)
├── data/                       # Local JSON storage (emissions.json, settings.json)
└── pages/                      # Isolated Application Modules
    ├── 1_Dashboard.py          # Visual Analytics & Report Generation
    ├── 2_Data_Entry.py         # Dynamic UI Ingestion Pipeline
    ├── 3_AI_Insights.py        # Autonomous Agent Interface
    └── 4_Settings.py           # Global Enterprise Context & Geofencing

⚙️ Installation & Setup

Prerequisites

• Python 3.12 or higher
• uv package manager (recommended)

1. Clone the repository

git clone https://github.com/yourusername/greenops.git
cd greenops

2. Install dependencies

Using uv (recommended):

uv sync

Or using standard pip:

pip install -r requirements.txt

Dependencies include: streamlit, pandas, plotly, python-dotenv, fpdf2, crewai[litellm], uuid

3. Configure Environment Variables

Create a .env file in the root directory and add your Groq API key for the AI engine:

GROQ_API_KEY=your_api_key_here

4. Boot the Application

streamlit run Home.py

---

🗄️ Database Structure

Local persistence utilizes flat JSON files mapped to Pandas DataFrames for rapid SME deployment.

Example emissions.json Schema

[
  {
    "date": "2026-04-04",
    "business_unit": "Main Office",
    "scope": "Scope 2",
    "category": "Electricity",
    "activity": "India Grid",
    "country": "India",
    "facility": "Ahmedabad HQ",
    "responsible_person": "Data Officer",
    "quantity": 1500.0,
    "unit": "kWh",
    "emission_factor": 0.82,
    "emissions_kgCO2e": 1230.0
  }
]

---

Acknowledgments

This project was developed starting from an initial architectural reference by AI Anytime. It has since been >95% rewritten, modularized, and enhanced with enterprise-grade ESG physics, cryptographic data guards, and FPDF reporting capabilities.