BackGovernment Civic Tech & AI Assistant

#04★ FEATUREDLive on Render

Yojana.AI

Intelligent citizen assistant matching Indian families with 50+ welfare schemes

AIPythonReact

LIVE CODE MAP ARCH MAP

Back to ProjectsGovernment Civic Tech & AI Assistant

BUILD #04★ FEATUREDLive on Render

Yojana.AI

Intelligent citizen assistant matching Indian families with 50+ welfare schemes

AIPythonReact

OPEN LIVE DEMO GITHUB GITDIAGRAM VIEW ARCHITECTURE ↓

ENGINEERING // ARCHITECTURE

How It Connects

Yojana.AI simplifies the complex landscape of 50+ central and state government welfare schemes. Citizens answer conversational prompts regarding income, state, caste, profession, land ownership, and age. An AI classification and rule engine evaluates eligibility matrices, predicts eligible grant amounts, and outputs an actionable application roadmap with required documentation checklists and direct government portal links.

Yojana.AI / ARCH MAP

LIVE

ARCHITECTURE ENGINE

INITIALIZING REPO...

SCROLL TO PANMERMAID.INK

GitDiagram Repository

Components

Conversational Questionnaire

React, Tailwind CSS

Scikit-Learn Scheme Classifier

Python, Scikit-learn

Scheme Knowledge Base

JSON Knowledge Base

Roadmap PDF Generator

PDFKit, Node.js

Open GitDiagram View Source

Yojana.AI / ARCH MAP

LIVE

ARCHITECTURE ENGINE

INITIALIZING REPO...

SCROLL TO PANMERMAID.INK

GitDiagram GitHub

System Components

EXECUTION PIPELINE

How It Works

01. Citizen

Citizen Profile Questionnaire

Answers straightforward questions regarding age, gender, occupation, family income, and residence.

→ Citizen Attribute Profile

02. Preprocessor

Feature Vectorization

Encodes categorical demographics into normalized numerical feature vectors.

→ Model Input Vector

03. Decision Tree Model

Eligibility Evaluation

Evaluates citizen vectors against 50+ government scheme qualification criteria.

→ Eligible Schemes Array

04. Yojana AI Core

Roadmap & Document Checklist

Cross-references required certificates (Caste, Domicile, Bank Account) for each qualifying scheme.

→ Personalized Application Dossier

Citizen

Citizen Profile Questionnaire

Answers straightforward questions regarding age, gender, occupation, family income, and residence.

→ Citizen Attribute Profile

Preprocessor

Feature Vectorization

Encodes categorical demographics into normalized numerical feature vectors.

→ Model Input Vector

Decision Tree Model

Eligibility Evaluation

Evaluates citizen vectors against 50+ government scheme qualification criteria.

→ Eligible Schemes Array

Yojana AI Core

Roadmap & Document Checklist

Cross-references required certificates (Caste, Domicile, Bank Account) for each qualifying scheme.

→ Personalized Application Dossier

STACK // ARCHITECTURE INTEL⚡ LIVE SPECS

Technical Breakdown

Granular architectural layers, runtime dependencies, and audited production decisions.

STACK: ML, Next.js, Scikit-learn

React with clean accessible wizard steps and instant score animations
copy

Node.js and Python microservice architecture
copy

Engineering Tradeoffs & Key Decisions

Why this architecture over alternatives

1 ARCHITECTURAL CHOICES

DECISION #1AUDITED

Decision Tree Model over Black-Box Neural Networks

↳

Selected interpretable Decision Trees for scheme qualification.

Government eligibility requires 100% auditable, explainable decisions so citizens know exactly why they qualify.

PRODUCTION TESTEDZERO REGRESSIONS

✓

Audited Production Outcomes & Metrics

HIGH IMPACT RESULTS

▸Developed an intelligent assistant to match users with eligible government schemes.

▸Built a classification model using Decision Trees to predict eligibility for 50+ schemes.

▸Integrated a user-friendly frontend to generate customized PDF application roadmaps.