Projects
HolisticMatch - Full-Stack Professional Directory Platform
Developed a full-stack discovery platform connecting clients with holistic therapy professionals through searchable listings and direct contact. Built React + TypeScript frontend with responsive grid layouts, advanced search filters, real-time state management, and Framer Motion spring physics animations. Implemented Django REST Framework backend with JWT authentication, email verification, role-based permissions, and 2-step registration form validation. Deployed AWS infrastructure with EC2 (Gunicorn + Nginx), S3 media storage, PostgreSQL on Supabase, and CloudFront CDN. Created AI-powered content moderation using AWS Rekognition for image validation (nudity/violence detection) and OpenAI API for text moderation with Portuguese regex fallback. Engineered secure authentication with JWT tokens, password reset via email, timing attack protection, and CSRF guards. Achieved 183/183 passing tests with pytest coverage, CI/CD via GitHub Actions, and SSL/TLS with Let's Encrypt.
Smart Gallery - AI Photo Management
Developed a full-stack serverless photo management application with AI-powered automatic tagging and real-time analytics. Built modern Pinterest-style gallery interface with Vue.js 3 Composition API and TypeScript, featuring drag-and-drop uploads, lightbox modals, and responsive design. Implemented comprehensive AWS serverless architecture using Lambda functions for image processing, Amazon Rekognition for intelligent photo classification, S3 for scalable storage, DynamoDB for metadata management, and CloudFront for global content delivery. Created real-time analytics dashboard with Chart.js visualizations, event-driven UI updates, and dark theme components.
Babi Santana - Professional Spanish Teacher Website
Developed a high-performance landing page for Spanish language education services with premium scroll-based animations. Built React + TypeScript frontend with GSAP ScrollTrigger for complex multi-phase hero parallax (5-layer composition with independent velocities), scrub-controlled section animations, and reversible entrance/exit effects. Implemented TailwindCSS mobile-first responsive design with custom color system and reusable Section component architecture. Created sophisticated animation sequences including zigzag layouts, auto-rotating testimonials carousel, elastic highlight effects on keywords, and magnetic CTA button with pulse animations. Engineered sticky pill navigation with hamburger menu, smooth hash-based section scrolling, and hover effects using GSAP timelines. Achieved 60fps performance with GPU-accelerated animations, centralized content management system, and optimized Vite build. Deployed to Vercel with automatic CI/CD pipeline.
NotionCards Dashboard
Developed an integrated dashboard system connected to the Notion API with role-based authentication. Implemented
an admin panel for full CRUD operations and a client panel for Notion card approval and visualization. Designed a
responsive interface using React and TailwindCSS, with seamless data synchronization via Notion API.
Test
credentials:
teste@teste.com / teste123
LFA Database
Built a modern web application to query fighters and display their participation across LFA events. Implemented automated web scraping using Playwright for robust data extraction and generated a comprehensive SQLite database with fighter statistics and event history. Built an interactive interface using Vite, React, and TypeScript with advanced filtering and search capabilities.
3D Interactive Hotsite
Interactive 3D showcase of the Lancia Fulvia HF built with Three.js, React and TypeScript. Features optimized 3D model, smooth camera controls, technical specifications menu, intelligent fallbacks and performance monitoring. Automated deployment via Vercel with S3 assets. Includes Vite build system, WebGL rendering, Vitest testing suite, and WCAG accessibility compliance.
SpaceX Falcon 9 Landing Prediction
Implemented a machine learning pipeline to predict Falcon 9 first-stage landing success, optimizing model selection for cost estimation and mission reliability. Applied data preprocessing, feature engineering, and hyperparameter tuning for Logistic Regression, SVM, Decision Tree, and KNN classifiers.
UFC Prospect Detection – Ensemble ML System
Developed a machine learning ensemble system that combines multiple anomaly and classification algorithms to identify potential UFC fighters. Integrated models including LOF, Logistic Regression, Isolation Forest, One-Class SVM, PCA + LOF, and Ensemble LOF, generating a consensus through weighted calibration and false-positive reduction. Achieved 80% real-world accuracy (12 of 15 verified prospects), with confirmed UFC and DWCS signings.
HcChat – Intelligent Chat Assistant with RAG (WIP)
Built a retrieval-augmented chatbot for HcUnit, combining Next.js 14, React 18, and FastAPI. Integrated DeepSeek for natural conversational AI and Pinecone for semantic knowledge retrieval. Deployed entirely on Render, with over 90% test coverage and production-ready CI/CD configuration.
Farm Jockey – Retro Animal Racing Game
Developed a complete 2D racing game with gambling mechanics using Python and Pygame, featuring procedural sprite generation, real-time physics simulation, and dynamic probability algorithms. Implemented MVC architecture with modular components, WebAssembly deployment via Pygbag, comprehensive unit testing with pytest, and CI/CD automation through GitHub Actions. Includes asset management system, state machine pattern, audio integration, and cross-platform compatibility with 60 FPS performance optimization. Deployed using AWS S3 for static hosting and CloudFront for global content delivery.
Blender Addons
Developed professional Python addons for Blender 3D to enhance workflow efficiency for 3D artists and animators. Tools include real camera utilities, array and duplication enhancements, render preset management, and integration with SynthEyes scripts for camera tracking.