Available for opportunities
Hi, I'm Joao.
I build things for the web.
Full Stack Developer passionate about crafting modern web applications with clean code and great user experiences.
Full Stack Developer passionate about crafting modern web applications with clean code and great user experiences.
Introduction
Hi, I'm Joao! I'm a Full Stack Developer passionate about coding and tech in general. I love building web applications and exploring new technologies that push boundaries.
Currently working as a Full Stack Developer Intern at Clauger, where I develop and maintain web applications using Angular, .NET, and more. I'm always eager to learn and take on challenging projects.
Developing and maintaining internal web applications — transport management, cybersecurity education platform, data automation, and more.
Two-year technical degree in Software Solutions and Business Applications. Full-stack web apps, databases, cybersecurity, and project management.
General Baccalauréat with a focus on Computer Science and Global English Studies (European section).
Full-stack transport & freight management system replacing a legacy PHP form. Complete request workflow with carrier coordination, cost allocation, and document generation.
Clauger — International industrial refrigeration and HVAC company operating across 20+ countries.
Role: Full Stack Developer (Internship) · Solo developer with supervisor validation · Oct 2024 - Dec 2025
The logistics department relied on an outdated PHP form to manage transport requests. This system was slow, lacked validation, had no tracking capabilities, and couldn't handle multi-site freight coordination. The company needed a modern, centralized platform for creating, tracking, and managing transport requests with carrier coordination, cost allocation, and document generation.
Client-server: Angular SPA ↔ .NET Core REST API ↔ SQL Server. Azure Entra ID for authentication. Docker containers for deployment. Swagger for API documentation.
IoT smart building control app. Real-time temperature, lighting, occupancy, and window monitoring via industrial automates connected through an OPC service chain.
Clauger — International industrial refrigeration and HVAC company.
Role: Full Stack Developer (Internship) · Solo developer (OPC/automate layer pre-configured by infrastructure team) · Jan 2025 - Feb 2025
Employees had no way to directly control their office environment without contacting facilities. The company wanted a self-service web interface connected to existing industrial automates (PLCs) via OPC protocol, letting users adjust temperature and lighting from their browser.
Three-tier: Industrial automates ↔ OPC Server ↔ SQL Server ↔ .NET REST API ↔ Angular SPA. The database acts as a shared communication layer between the OPC world and the web application.
Full-stack expense report management app for the fictional pharmaceutical company Galaxy Swiss Bourdin (BTS SIO exam). Angular frontend, Node.js backend, MongoDB, deployed on AWS.
GSB (Galaxy Swiss Bourdin) — Fictional — Standard BTS SIO exam scenario. Pharmaceutical company whose sales representatives submit expense reports for reimbursement.
Role: Solo Developer (BTS SIO exam project) · Jan 2025 - Mar 2025
GSB sales representatives travel frequently and submit expense reports. The company needs a web application allowing employees to create and track expense reports, and administrators to review, validate, or reject them with role-based access and business rule validation.
Decoupled client-server: Angular SPA ↔ Node.js/Express REST API ↔ MongoDB. JWT authentication. Deployed on AWS EC2. Two separate Git repositories.
WinForms desktop app for managing medical prescriptions, patient records, and pharmaceutical databases. SHA2-256 auth, PDF export with iText7, Docker.
GSB (Galaxy Swiss Bourdin) — Fictional — Standard BTS SIO exam scenario. Doctors prescribe medications; the system manages prescriptions, patients, and the pharmaceutical catalog.
Role: Solo Developer (BTS SIO exam project) · Nov 2024 - Jan 2025
GSB doctors need a desktop application to manage medical prescriptions efficiently: patient records, pharmaceutical database, prescription creation with dosage management, secure authentication, and PDF export for printing and archival.
Two-tier desktop: WinForms UI ↔ MySQL database. Business logic in the application layer with repository pattern. Docker-containerized MySQL instance.
Full-stack fitness app combining workout tracking, macro nutrition monitoring, progression analytics, and social networking. Live at stayraw.fr.
Personal Project — Self-initiated to solve a personal need for a unified fitness tracking platform.
Role: Solo Full Stack Developer · Mar 2025 - Present
Existing fitness apps either focus on workouts OR nutrition, rarely both. I wanted a single platform to log workouts with progression tracking, monitor daily macros, calculate rep maxes, and share progress with friends.
Serverless: Next.js (Vercel) ↔ Supabase (PostgreSQL + Auth + Realtime + Storage). API routes for server-side logic. Supabase handles authentication, database, and real-time subscriptions.
Interactive Ubuntu terminal portfolio with working CLI, draggable app windows, GNOME-styled ecosystem, and easter eggs. You might be viewing it right now!
Personal Project — Creative portfolio to showcase technical skills through an interactive Ubuntu desktop simulation.
Role: Solo Developer · Nov 2025 - Present
Standard portfolio websites are generic and forgettable. I wanted a portfolio that itself demonstrates technical ability — a pixel-perfect Ubuntu desktop with working terminal, draggable windows, a music player, and a full app ecosystem. The portfolio IS the project.
LLaMA, Mistral, Gemma: how open-source is reshaping the AI landscape.
In 2025, open-source language models reached performance levels that rival proprietary solutions. Meta with LLaMA 3, Mistral AI with its Mixtral models, and Google with Gemma have significantly closed the quality gap with GPT-4 and Claude.
The Hugging Face ecosystem plays a central role in this democratization. With over 500,000 hosted models, the platform enables developers to fine-tune and deploy models tailored to their specific needs without relying on expensive proprietary APIs.
This trend has a direct impact on software development: companies can now integrate AI capabilities into their products while retaining control over their data and reducing infrastructure costs.
Claude Code, Copilot, Devin: the emergence of agents that can code autonomously.
Autonomous AI agents represent a major evolution from simple chat assistants. Unlike a classic LLM that answers a question, an agent can plan, execute actions, use tools, and iterate on its results to accomplish complex tasks.
Anthropic's Claude Code illustrates this trend well: it can navigate a codebase, modify files, run tests, and fix bugs autonomously. GitHub Copilot is evolving in the same direction with its agent mode, while Cognition AI's Devin positions itself as a complete 'AI software engineer'.
The ReAct (Reasoning + Acting) pattern is at the heart of these systems: the agent reasons about the task, chooses an action, observes the result, then adjusts its strategy.
IoT + ML to anticipate industrial failures — directly tied to my experience at Clauger.
Predictive maintenance combines IoT sensors and machine learning algorithms to anticipate industrial equipment failures before they occur. In the industrial refrigeration sector, where I work at Clauger, this approach is particularly relevant.
Compressors, heat exchangers, and refrigeration systems continuously generate temperature, pressure, and vibration data. ML models analyze these streams to detect abnormal patterns and predict failures days in advance.
The economic impact is considerable: reducing unplanned downtime by 30 to 50%, extending equipment lifespan, and optimizing maintenance interventions.
The European legal framework for AI takes effect — implications for developers.
The European AI Act, progressively coming into force since 2024, establishes the world's first comprehensive legal framework for artificial intelligence. Based on a risk-level approach, it classifies AI systems into four categories: unacceptable risk (banned), high risk, limited risk, and minimal risk.
For developers, the implications are concrete: high-risk AI systems must meet strict requirements for transparency, technical documentation, training data quality, and human oversight.
Foundation models (GPT, Claude, LLaMA) are subject to specific transparency obligations: documentation of capabilities, copyright compliance in training data, and labeling of AI-generated content.
