How naman-56-56/saarthi works
Project Type & Purpose
• This is a full-stack web platform (“Saarthi”) combining:
• Next.js (React + TailwindCSS) frontend under /frontend/
• Django REST API backend with an AI/ML recommendation engine under saarthi/
• PostgreSQL + Redis for data, external integrations for internship data, and containerized deployment
Key Architectural Components
a. User Layer (Students, Employers, Admins)
b. Frontend Layer (Next.js App Router, UI Components, Hooks, Utilities)
c. Backend Layer (Django REST Framework, Authentication, Business Logic)
d. ML Recommendation Engine (Text preprocessing, vectorization, filtering algorithms)
e. Data Layer (PostgreSQL, Redis cache, Cloud storage for media)
f. External Integrations (Internshala API, PM Portal data, Kaggle datasets)
g. Infrastructure (Docker, Kubernetes, NGINX reverse proxy/CDN)
Relationships & Data Flows
• Web Browser → Frontend (HTTPS)
• Frontend → Django API (REST calls)
• API → ML Engine (in-process calls)
• API → Database / Redis / Storage (SQL, cache, file APIs)
• External Data Sources → API / ML Engine (scheduled or on-demand fetch)
• Infrastructure layers (Load Balancer, Containers, CDN) wrap the entire circle
Architectural Patterns & Principles
• Monolithic codebase with microservices-inspired separation of concerns
• Layered architecture: Presentation → API → Business Logic → Data
• Modular React components + custom hooks for UI
• Django apps for users and recommendations, clear service boundaries
• ML engine as an internal service (scikit-learn models loaded via joblib)
• Containerization & orchestration for portability and scalability