
Focused on backend systems, cloud architecture, and building scalable applications.
Developed AR indoor navigation Android prototype with extensible Python backend architecture. Delivered working MVP in 4 weeks with voice guidance and multi-floor navigation capabilities.
INESC TEC →Published Spring Boot authentication library to Maven Central with comprehensive CI/CD pipeline, automated testing, and security scanning.
Created and Optimized Match Dinner Mondays platform, reducing image sizes by 95% and achieving sub-second load times through advanced compression techniques.
Completed Harvard's AI course, focusing on machine learning algorithms and AI system architecture.
View Certificate →Transitioned focus to backend systems, API design, and database optimization through hands-on projects.
Completed Harvard's Introduction to Computer Science, establishing strong foundation in algorithms and data structures.
View Certificate →Started with system-level programming, inspired by performance-critical applications.

Published Spring Boot authentication library on Maven Central with zero-code integration. • CSRF protection and Redis-backed rate limiting • JWT security with OWASP compliance • Only 12ms request overhead • Benchmarked with 100k users: - 22% query improvement - 73x faster pagination • Full CI/CD with automated testing

Reinforcement Learning pipeline built from scratch using Monte Carlo Tree Search (MCTS) and dual-head ResNets. • Architected a custom RL pipeline using MCTS and PyTorch • Bootstrapped policy networks using Knowledge Distillation from Stockfish • Implemented dynamic batch sizing and mixed-precision (AMP) for GPU efficiency • Visualized real-time search tree expansion and node evaluation • Achieved competitive play vs standard engines

AI-powered resume optimization using NVIDIA's Llama-3.1 Nemotron Ultra (253B parameters). • Engineered prompt optimization for consistent JSON responses • Reduced token consumption with structured scoring • 4 comprehensive evaluation criteria • Advanced error handling for reliable AI feedback • Optimized for professional resume enhancement

High-performance stock portfolio tracker with advanced caching and API optimization. • Redis caching: 20+ seconds → sub-second load times • Parallel API processing across 4 financial data providers • Intelligent fallback strategies for data consistency • Real-time portfolio tracking and analytics • Optimized for high-frequency data updates

Performance-optimized dating platform with advanced optimization techniques. • WebP compression: 3-5MB → ~200KB images • Lazy loading for improved mobile experience • SEO strategy achieving first-page Google rankings • PWA functionality for enhanced user engagement • Direct client collaboration with iterative development

Full-stack blog platform with comprehensive real-time features. • Real-time updates and live content synchronization • User authentication and authorization system • Complete CRUD operations for content management • Firebase backend services integration • Responsive design for all devices
Mentor testimonial from my latest internship
My overall assessment is very positive: he is a young engineer with excellent potential, autonomy, and technical maturity above the average for his level, ready to join international teams and demanding environments.
Prof. Fernando Cassola Marques
Phd researcher INESC TEC / Invited Professor FEUP · Porto, 15 October 2025
fernando.c.marques@inesctec.ptFeel free to reach out for collaborations or just a friendly hello.
Porto, Portugal