Votum: Real-Time Political Campaign Intelligence Platform

The Challenge: Political campaigns generate massive amounts of data, but turning that noise into actionable strategy is slow and expensive. The goal was to build a "war room" SaaS that could ingest real-time news, analyze voter sentiment via AI, and automate communication strategies instantly.

The Solution: I architected and built Votum, a comprehensive intelligence platform that integrates Big Data with Generative AI. It allows political strategists to simulate scenarios, track viral trends in real-time, and generate compliance-checked content across multiple channels.

Real-Time Trend Monitoring (Votum Sensus): The system scrapes and analyzes news to identify "Top 5 Daily Topics," scoring them by Relevance and Virality using NLP models.

Geospatial Intelligence: Integrated a Python/Streamlit module directly into the React frontend to render heavy electoral data maps (heatmap of Minas Gerais), allowing for granular regional analysis.

AI Simulation Engine (The Oracle): Built a predictive module using LLMs to simulate voter reactions to specific narratives before they go public, acting as a digital focus group.

Automated Content Pipeline: A module that automatically converts strategic insights into ready-to-post scripts for TV, Radio, and WhatsApp.

Under the Hood (Tech Stack) This is not just a wrapper; it’s a complex distributed system:

Frontend: React 18 (MUI, Zustand) for a responsive, app-like experience.

Backend: Node.js/Express handling complex logic with Redis (BullMQ) for asynchronous job processing.

Data: Google BigQuery for heavy lifting and PostgreSQL for transactional data.

Real-Time: WebSockets for live updates on trend tracking and generation progress.

Methodology: AI-First Engineering ("Vibe Coding") What makes this project unique is the delivery speed. I utilized an AI-Accelerated Development workflow. By orchestrating LLMs to assist in generating boilerplate code, complex SQL queries, and React components, I acted as the Systems Architect, reducing a typical 6-month development cycle into weeks without sacrificing code quality or architectural robustness.