An AI-Driven SaaS Platform for Automated Portfolio Generation, Live Publishing, and Schema-Guided Template Rendering

Abstract

This paper presents Portiva, a cloud-based Software-as-a-Service (SaaS) platform that automates the creation, live publishing, and management of professional portfolio websites. The system accepts structured user data through a guided multi-step form, delegates content synthesis to a locally-hosted Large Language Model (LLM) via Ollama, and renders a fully live portfolio on a unique public URL — all within minutes of signup. The core contribution of this work is a novel schema-driven placeholder mapping architecture in which each template declaratively specifies its data requirements through a structured placeholder_schema stored in the database. The AI engine reads this schema to generate precisely the content each template needs, eliminating hardcoded field mappings and enabling templates to be added without modifying application code. The platform is built on a modern stack — React 19, Spring Boot 3.2, FastAPI, PostgreSQL, and Redis — and follows production-grade practices including asynchronous AI generation, Redis-based result caching, soft deletes, and Razorpay-integrated subscription management. Experimental evaluation demonstrates sub-200ms API response for portfolio generation, successful dynamic template selection through a scoring algorithm, and complete end-to-end portfolio delivery in under three minutes on commodity hardware.