Back to Shubham's profile
8views

ESGForge

An AI-powered ESG platform that streamlines BRSR reporting through intelligent document processing, analytics, validation and data traceability.

Shubham Kumar Behera
  • Problem, approach and result explained
  • Expert reviewed· 6 reviews
ESGForge

—

6

—

7

—

6

Overview

The problem

Organizations face difficulties collecting, validating, monitoring and consolidating ESG data across multiple projects, business units and subsidiaries. Electricity bills and other supporting documents are often processed manually, making it difficult to identify individual invoices, extract accurate consumption figures, calculate carbon emissions and maintain reliable evidence.

Existing reporting workflows also lack transparent data traceability and structured approval processes. This can lead to inconsistent data, delayed reporting, calculation errors and reduced confidence in BRSR disclosures. Our objective was to develop ESGForge, an integrated ESG intelligence and BRSR reporting platform that simplifies ESG data management, improves data quality, automates document processing and makes every reported figure traceable to its supporting evidence.

The approach

We developed ESGForge using React, Vite, FastAPI, Python and PostgreSQL/Supabase, with an architecture designed around ESG data collection, validation, analytics, evidence management and BRSR reporting. Our approach combines:

Role-Based Management: Separate workspaces for Group Admin, Subsidiary Management, Project Contributor and other operational roles, with a structured multi-level approval workflow planned for project and business-unit management.

BRSR Reporting: Structured disclosure management covering Sections A, B and C, including Principles 1–9 and their Essential and Leadership Indicators.

AI/ML Intelligence: Prototype modules for anomaly detection, data quality assessment, ESG scoring, prediction, benchmarking and materiality analysis.

Billing Document Intelligence: A document-processing workflow designed to identify multiple electricity invoices within uploaded PDFs, extract invoice information, organize monthly consumption and calculate electricity-related emissions.

Digital Traceability: Connecting ESG values with their source records, supporting documents, validation results, calculations and approval history.

BRSR Readiness: Monitoring reporting completeness, missing information, evidence status and validation requirements.

For electricity emissions, purchased grid electricity is classified under Scope 2. Emissions are calculated using electricity consumption and an appropriate, verified emission factor.

The result

We developed a working ESGForge frontend with dashboards, role-based navigation, energy consumption records, submission workflows, evidence management, analytics and BRSR reporting interfaces.

The analytics prototype includes six modules: data quality scoring, anomaly detection, prediction, ESG scoring, benchmarking and materiality analysis. It generates seven categories of analytical outputs, and anomaly detection successfully identified deliberately injected test anomalies.

The billing documentation prototype demonstrates PDF processing, invoice extraction, monthly electricity aggregation, financial summaries, Scope classification and carbon-emission calculations.

The platform also includes BRSR reporting and readiness interfaces designed to improve disclosure tracking and data traceability.

Some capabilities remain partially implemented. In particular, scanned-document OCR accuracy, complete end-to-end database integration, verified emission-factor configuration and the expanded multi-level approval workflow require further testing and development.

Reflection

Building ESGForge helped us understand that ESG reporting is not simply about collecting numbers; it requires reliable evidence, accurate calculations, accountability and transparent approval processes.

One of our biggest challenges was integrating document extraction, data validation, analytics and reporting into a single workflow while maintaining consistency across organizational levels.

We learned that AI-generated results must be validated, emission factors must come from reliable sources, and every reported value should be traceable to its original evidence.

Our next priorities are improving OCR accuracy for scanned and multi-invoice PDFs, completing the Project Admin and Business Unit Management approval hierarchy, strengthening database integration, and conducting comprehensive end-to-end testing.

Our long-term vision is to make ESGForge a reliable ESG intelligence and BRSR reporting platform that helps organizations move from manual reporting towards evidence-driven, transparent and intelligent sustainability management.

Submitted to The BRSR Reporting Portal

Your idea in one line

ESGForge transforms project-level ESG data into audit-ready BRSR reporting through verified carbon calculations, evidence-linked traceability, and real-time change-impact analysis—from individual project to the entire group.

What will be working by the finale?

By the Grand Finale, we aim to deliver an integrated ESGForge workflow that supports: 1. Project-level data entry for Scope 1 diesel consumption, Scope 2 purchased electricity, and Scope 3 upstream material transportation. 2. A documented emission-factor register and unit-aware CO₂e calculations. 3. Supporting-document uploads, evidence linking, data validation, and submission and approval workflows. 4. Emissions aggregation from project to business unit, subsidiary, and group level. 5. Generation of BRSR Section C reports from stored and validated data. 6. Digital Traceability and the Data Passport to trace reported values to their source activities, calculations, evidence, and approvals. 7. A Readiness Engine that identifies missing data, evidence, validation, and approval requirements. 8. Integrated testing of the complete workflow, with analytics supporting ESG data-quality assessment and decision-making.

Tech stack and hardware you plan to use

Frontend: React, TypeScript, Tailwind CSS. Backend: Python, FastAPI, REST APIs, and JSON. Database: PostgreSQL using Supabase. Data processing and AI/ML: Python, Pandas, NumPy, and scikit-learn for data preprocessing, validation support, anomaly detection, prediction, ESG scoring, benchmarking, and materiality analysis. Core functionality: Scope 1/2/3 activity-data management, emission-factor mapping, unit-aware CO₂e calculations, evidence management, organizational aggregation, BRSR Section C reporting, digital traceability, and readiness assessment. Security: JWT-based authentication and role-based access control (RBAC), subject to implementation and testing. Development tools: Git, GitHub, VS Code, and a web browser. Hardware: A standard development laptop or computer and an internet connection. No specialized hardware is required.

A diagram, sketch or mock-up of your idea
A diagram, sketch or mock-up of your idea: ESGForge
What have you built or tested so far?

https://aavishkara.vercel.app/

Pitch deck (PDF)
Pitch deck (PDF)PDF
What's working now?

Working: ESGForge has a functional React-based dashboard with five role-based interfaces, organization and project management, energy consumption records, evidence management, BRSR reporting, KPI calculations, analytics, anomaly detection, data quality intelligence, traceability and reporting. The frontend supports role-specific access and demonstrates the ESG data management workflow. Multi-level approval workflows, automated electricity bill processing, OCR-based invoice extraction, Scope classification, carbon emission calculations and BRSR readiness are under improvement. Some ML and analytics modules have been prototyped and require further end-to-end integration and validation. Partially Completed: The complete production-ready workflow with all five approval levels, reliable OCR for multiple invoices in scanned PDFs, fully verified emission factors, and complete backend/database integration and regression testing is not yet finished. Our focus is to improve accuracy, traceability, evidence verification and reliable ESG reporting while preserving the existing working features.

Deck or report (PDF)
Deck or report (PDF)PDF

AI tools used

ClaudeGitHub CopilotPerplexityChatGPT

ChatGPT for project planning, system architecture, technical guidance and debugging. Claude for code analysis, frontend development and implementation planning. GitHub Copilot for coding assistance and Perplexity for technical research and reference discovery.

Links & files

Artifacts

1

Gallery

5