Junior Data Scientist | Clinical AI & Health Data Science | Machine Learning | MSc Physiotherapy | Clinical Research | Digital Health |
For the past 10 years, I worked with human health data in real clinical environments — interpreting heart rates, respiratory capacity, recovery signals and post-COVID complications as a physiotherapist specialized in cardiovascular rehabilitation.
That experience taught me that healthcare data is never just numbers: it represents human variability, uncertainty and critical decisions that directly impact people's lives.
Today, I combine clinical expertise, scientific research and machine learning to build AI systems that transform complex healthcare data into actionable and clinically meaningful solutions.
My work focuses on bridging healthcare knowledge and AI engineering through predictive modeling, interpretable machine learning, health analytics and scalable data-driven applications.
- 🎓 MSc in Physiotherapy — Universidade Federal de Pernambuco (UFPE)
- 🏥 Specialization in Cardiovascular Rehabilitation
- 🚑 Clinical experience at a Cardiorespiratory Emergency Hospital (Brazil)
- 🔬 Research in post-COVID cardiorespiratory, musculoskeletal and vascular repercussions
- 📝 Publications presented at the American Thoracic Society (ATS)
- 🆔 ORCID: 0000-0002-3269-0654
| Area | Tools | Level |
|---|---|---|
| Data Analysis | Python, Pandas, NumPy | ⬛⬛⬛⬜⬜ Building |
| Visualization | Matplotlib, Seaborn, Plotly | ⬛⬛⬛⬜⬜ Building |
| Machine Learning | Scikit-learn, XGBoost, Optuna | ⬛⬛⬛⬜⬜ Building |
| Deep Learning | CNN, NLP, Computer Vision, LLMs | ⬛⬛⬜⬜⬜ Building |
| Explainability | SHAP | ⬛⬛⬜⬜⬜ Learning |
| Databases | SQL, MySQL, PostgreSQL | ⬛⬛⬛⬜⬜ Intermediate |
| App Deployment | Streamlit, FastAPI, Docker | ⬛⬛⬜⬜⬜ Learning |
| Clinical Statistics | SPSS, Excel | ⬛⬛⬛⬛⬜ Experienced |
| Project | Description | Stack | Status |
|---|---|---|---|
| 🧬 ML – AIGenix: Antigen Predictor | Collaborative AI project developed during Saturdays.AI Madrid. Protein antigenicity classifier from FASTA sequences. Predicts which pathogen proteins are most likely to be recognised by the human immune system. End-to-end pipeline with feature engineering from amino acid sequences + Random Forest + Streamlit app | Python · Scikit-learn · Biopython · Streamlit · Joblib | 🔄 In progress |
| 🧠 ARIA – AI Stroke Risk Analysis | Hybrid AI pipeline for stroke risk prediction combining traditional machine learning and CNN-based deep learning models using clinical and lifestyle variables. Includes EDA, feature engineering, explainability analysis and healthcare-focused risk interpretation | Python · Scikit-learn · CNN · XGBoost · SHAP · Pandas · Streamlit | ✅ Complete |
| 🌸 ML – PelvIA: Urinary Incontinence Classification | Multiclass classification of female urinary incontinence type using NHANES dataset (CDC, 2017–2023). Soft voting ensemble (XGBoost + RF + LightGBM) with SMOTE balancing, Optuna tuning + Streamlit app deployed with Docker | Python · XGBoost · LightGBM · Optuna · SMOTE · Streamlit · Docker | ✅ Complete |
| 🫀 EDA – Heart Disease (UCI Statlog) | Exploratory analysis of clinical cardiac variables: univariate, bivariate, correlation analysis + storytelling report for clinical and technical audiences | Python · Pandas · Seaborn · Matplotlib | ✅ Complete |
| 🌍 ML – Life Expectancy Calculator (WHO) | Predicts life expectancy from WHO socioeconomic and health indicators. 4 regression models trained in parallel + interactive simulator combining predictions | Python · Scikit-learn · Joblib · Jupyter | ✅ Complete |
| Project | Description | Stack | Status |
|---|---|---|---|
| ⚡ MegaWattle | Collaborative project developed during the EuroGenAI Hackathon League for Social Good. AI-powered backend platform focused on scalable API architecture, automation workflows and production-oriented data processing. Includes FastAPI services, database integration and Dockerized deployment | Python · FastAPI · PostgreSQL · Docker · SQLAlchemy · REST APIs | ✅ Complete |
| 🏦 ML – Customer Subscription Prediction | End-to-end ML pipeline: EDA → SMOTE → XGBoost + Optuna tuning → SHAP explainability → Streamlit app. AUC-ROC: 0.792. No data leakage. Production-ready | Python · XGBoost · Optuna · SHAP · Streamlit · Pytest | ✅ Complete |
| 🏋️ GymPro – Backend REST API | Production-grade REST API for gym management. MVC architecture, JWT auth, CI/CD pipeline, full test suite and Docker deployment | Python · FastAPI · PostgreSQL · Docker · SQLAlchemy · Pytest | ✅ Complete |
| 🗄️ CRUD – Python + MySQL | Relational database operations with MVC architecture — educational module designed and taught to bootcamp peers | Python · MySQL · MVC | ✅ Complete |
├── 🫀 Cardiovascular risk prediction
├── 📡 Wearables & health sensors (smartwatch, ECG, PPG, HRV)
├── 🏋️ Exercise physiology + cardiac monitoring during rehabilitation
├── 🧬 Post-COVID health data (cardiorespiratory & vascular)
├── 🏥 Clinical decision support systems
└── 🌍 Public health & population data
- Factoría F5 — AI & Data Science Bootcamp (Module 3: Deep Learning)
- Saturdays.ai Madrid — 8th Edition (Applied AI projects)
Interested in building AI and data-driven systems that solve real-world healthcare problems through machine learning, clinical research and scalable technology.
Particularly motivated by:
- Clinical AI & decision support systems
- Preventive and predictive healthcare
- Wearables & physiological signal analysis
- Cardiovascular and rehabilitation technologies
- Biomedical and life science AI
- Healthcare data platforms & intelligent APIs
I bring a combination of healthcare expertise, scientific research experience and applied machine learning focused on transforming complex clinical data into actionable and clinically meaningful solutions.
My goal is to contribute to AI systems that are not only technically robust, but also interpretable, evidence-based and designed around real human needs.
📍 Based in Madrid, Spain | 🗣️ PT · EN · ES | 📫 LinkedIn
"The best data scientists in health are not those who know the most algorithms, but those who know what questions are worth asking."