Verdex/Docs

Technical Architecture

Deep dive into the Verdex platform architecture, AI engines, and knowledge base.

System Overview

Verdex is built as a modern web application with an AI-powered backend that processes project documents, validates compliance, and generates recommendations.

Next.js
Frontend
Groq & ASI:One
AI Engine
Pinecone
Vector DB
RAG
Knowledge

Frontend Layer

The frontend is built with Next.js and provides four main interfaces:

Upload / Input
Project document upload and form input
Transition Assessment
LMA 5 Components scoring dashboard
Financing Pathway
DFI matching and structure recommendations
Results & Export
PDF report generation and clause search

AI Engine

The AI engine is powered by Groq (Llama 4 Maverick) with RAG (Retrieval-Augmented Generation) for accurate, grounded responses.

1
Green Draft Generator
Creates LMA-compliant documentation drafts from project data
2
Clause Matcher
Semantic search across indexed LMA documents using vector embeddings
3
KPI & SPT Generator
Recommends sector-specific KPIs based on SBTi pathways

Knowledge Base

20 indexed documents across LMA core documentation (15), Africa-specific resources (3), and external standards (2) including SBTi Net-Zero and Paris Agreement articles. Full specifications in Knowledge Base documentation.

Tech Stack

json
{
  "frontend": {
    "framework": "Next.js 16",
    "language": "TypeScript",
    "styling": "Tailwind CSS",
    "icons": "Lucide React"
  },
  "ai": {
    "llm": "Groq (Llama 4 Maverick)",
    "embeddings": "BGE-small-en",
    "vectorDb": "Pinecone"
  },
  "features": {
    "pdfParsing": "pdf-parse",
    "pdfGeneration": "jsPDF + AutoTable",
    "animations": "GSAP"
  }
}

Scalability & Design

Stateless Architecture

No server-side sessions. Each request is independent, enabling horizontal scaling.

Cloud-Native

Deployed on Vercel (frontend) and Railway (embedding service). Auto-scales with demand.

Vector Database

Pinecone handles millions of embeddings. Clause search scales without index rebuilds.

API-First

All features accessible via REST APIs. Easy integration with existing bank systems.

💡From One to Thousands
Architecture handles one project or thousands without infrastructure changes. Same codebase, same validation logic, same compliance standards.
✅Built for LMA Edge Hackathon 2025
Verdex was developed for the LMA Edge Hackathon 2025, demonstrating how AI can bridge the climate finance gap in Africa through automated compliance validation.
Previous
Clause Library
Next
Knowledge Base