"In an era of accelerated AI, adaptability drives innovation."

I am a systems thinker who architects and engineers intelligent systems, inspired by the principles of nature and cognition. My approach is rooted in a deep academic foundation of the most successful systems to ever exist: those found in biology, and the human mind. I leverage first principles from these domains to engineer AI & ML systems and software that is not only powerful, but also robust, intuitive, and meaningful.

This is not just a job or a career for me; it is the practical application of a lifelong curiosity, and years of formal academic studies, into the nature of: life & biological systems, complex systems, philosophy, psychology, cognition & learning, languages & linguistics, programming & software development, data science, artificial intelligence & machine learning.

As we enter the era of agentic AI & physical AI (robotics), my mission is to move beyond simply studying, developing my own projects, or building tools, and begin orchestrating intelligent workforces in companies and institutions, with likeminded individuals. I am driven to architect and govern these powerful new systems responsibly and am looking for a team that shares this vision.

Optimized SOTA LLMs via RLHF	Scalable Multi-Agentic AI
AI for Ecological Research	Research Workflow Optimization

🏢 Professional Experience	🎓 Academic Background

Here is a comprehensive showcase of my projects, demonstrating my skills in building production-grade, scalable, and innovative AI systems from end to end across multiple domains.

A comprehensive multi-agent system built with modern async Python, showcasing Agent Communication Protocol (ACP) and Model Context Protocol (MCP) capabilities through a collaborative research workflow.

• Key Features: Engineered with a high-performance RabbitMQ message bus, containerized with Docker and Kubernetes, and fully observable with a Prometheus/Grafana stack.
• Technologies: Python, FastAPI, RabbitMQ, Docker, Kubernetes, ACP, MCP Explore the Architecture →

My winning project for FIAP's 2025.1 Global Solution Challenge. A visionary multi-agent platform designed to predict and manage large-scale events in Brazil by fusing Agentic AI with concepts from Brazilian folklore.

• Key Features: Five autonomous "Guardian" agents for different threat domains, with a fully functional MVP for fire risk prediction using real-time IoT sensor data.
• Technologies: Agentic AI, Python, FastAPI, Docker, MicroPython, ESP32, IoT See the Award-Winning Code →

A multi-agent AI platform for industrial IoT that predicts machine failures and automates maintenance scheduling, built entirely from scratch to ensure maximum performance and control.

• Key Features: Custom-built agentic architecture (no frameworks), leverages TimescaleDB for high-performance time-series data, and is fully containerized with Docker.
• Technologies: Python, FastAPI, PostgreSQL, TimescaleDB, Docker, Streamlit Check out the SaaS Platform →

A production-ready framework implementing three key post-training techniques to enhance and align Large Language Models: Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and Reinforcement Learning (GRPO).

• Key Features: Modular architecture, educational notebooks, YAML-based configuration, and built-in benchmarking tools based on the DeepLearning.AI course.
• Technologies: PyTorch, TRL, Hugging Face, Fine-Tuning, DPO, Reinforcement Learning Learn about LLM Alignment →

An AI-powered multi-agent system built with CrewAI that automatically reviews GitHub repositories, providing expert-level feedback on code structure, documentation, and best practices.

• Key Features: A team of specialized AI agents, a modern Streamlit web interface for real-time analysis, and a full CI/CD pipeline for quality assurance.
• Technologies: CrewAI, LangChain, Streamlit, Python, CI/CD, Docker View the Project →

An advanced automation platform designed to streamline digital communication. It uses a team of CrewAI agents to process emails, draft replies, and transform newsletters into social media content.

• Key Features: Intelligent email management, multi-platform social media automation, and a modern full-stack architecture.
• Technologies: React.js, FastAPI, PostgreSQL, CrewAI, Tailwind CSS, JWT See the Automation System →

An AI-powered system that automates invoice processing, drastically reducing manual effort.

• Key Features: Reduced processing time by over 85% and uses RAG with FAISS for intelligent error classification. Built with multiple frontend (React/Next.js) and deployment options.
• Technologies: Next.js, React, TypeScript, AWS, LangChain, Streamlit, RAG See the Full-Stack Solution →

A fully local, privacy-friendly RAG-powered chat application that runs entirely on your machine for secure document interaction.

• Key Features: Uses Google's Gemma model via Ollama for local LLM inference, FAISS for vector search, and a modern UI built with Reflex.
• Technologies: Reflex, LangChain, HuggingFace, FAISS, Ollama, RAG, Local-AI Explore the Privacy-First App →

An end-to-end machine learning platform to predict the Total Cost of Attendance for international higher education. A SuperDataScience Community Project.

• Key Features: Achieved a 96.44% R² score with an XGBoost Regressor, deployed via both a Streamlit web app and a FastAPI service, all containerized with Docker and automated with CI/CD.
• Technologies: Scikit-learn, XGBoost, MLflow, Streamlit, FastAPI, Docker, CI/CD Explore the EdTech Platform →

A deep learning solution that classifies 14 different crop diseases across four species (corn, potato, rice, wheat) from leaf images with high accuracy.

• Key Features: A Convolutional Neural Network (CNN) trained on over 13,000 images, deployed via a user-friendly Streamlit interface for real-time predictions.
• Technologies: Deep Learning, Computer Vision, CNN, TensorFlow, PyTorch, Streamlit See the Disease Detection Model →

A collection of high-performance Python tools for bioinformatics, including DNA sequence analysis, gene expression analysis, and a pipeline that uses ML to predict disease risk from genetic variants.

• Key Features: Combines population genetics with ML, features ORF detection, PCA for pattern recognition, and robust data processing.
• Technologies: Python, Bioinformatics, Genomics, PyTorch, Scikit-learn Explore Bio-AI Tools →

An advanced climate risk prediction system using ensemble machine learning and deep learning, delivered via a production-ready REST API.

• Key Features: Combines multiple ML models (XGBoost, LSTM) for robust forecasting and integrates real-time weather data for comprehensive analysis. Fully containerized and CI/CD ready.
• Technologies: Python, FastAPI, Ensemble ML, Deep Learning, Docker, CI/CD Check out the API →

A systematic exploration into decoding financial market patterns using ML, developed as a Scientific Initiation Project at UniAcademia.

• Key Features: Uses Random Forest classifiers to generate signals from technical indicators (RSI, MACD) and Elliott Wave Theory, with a robust backtesting engine to prevent lookahead bias.
• Technologies: Python, Scikit-learn, Quantitative-Finance, Algorithmic-Trading, Fintech Analyze the Trading System →

• 🤖 Agentic AI Portfolio: A collection of advanced multi-agent systems for automation, using frameworks like CrewAI, LangChain, and AutoGen. View →
• 📊 Machine Learning Portfolio: A comprehensive showcase of ML projects spanning supervised, unsupervised, and reinforcement learning techniques. View →

"The most flexible element is the one that controls the system."