Hire the Best Data Scientists
in Portugal

I'm a Data Analyst with a background in Industrial Engineering. I specialize in process optimization, data visualization, and leveraging data to address business challenges and enhance organizational efficiency. I've worked on automating data acquisition, analyzing market trends, and designing performance dashboards. My skills include Python, SQL, VBA, Power BI, and MS Excel. My focus is on applying analytical skills to drive efficiency and effectiveness. As a Master's in Data Science student, my goal is to help organizations make better decisions by uncovering insights in their data.

I build AI systems that turn business problems into measurable outcomes. Data Scientist and AI Engineer with 8+ years of experience across private and research sectors. I move between modern LLM systems and classical ML: RAG assistants, LLM agents, document extraction pipelines, and predictive models. What I build 🤖 RAG assistants & LLM agents Chat assistants over your documents that answer with citations you can check. I build the retrieval (hybrid search over Postgres/pgvector), the agent logic (LangGraph), and the evaluation (RAGAS golden sets, Langfuse tracing), so quality is measured and every change can be tested. 🎯 LLM pipelines & structured extraction Turn unstructured documents (reports, PDFs, clinical notes) into clean, validated, structured data ready for downstream analytics. ⚡ Predictive modeling & forecasting ML models for demand forecasting, risk estimation, and classification problems. 🔬 Computer vision & medical imaging AI Deep learning pipelines for image classification and clinical outcome prediction, including explainability outputs for clinical and research teams. 🛠️ End-to-end ML systems Full pipeline ownership from raw data to deployed model, ETL, feature engineering, model training, evaluation, and monitoring. Results 📌 Built a RAG assistant for contract clause research over 40 SEC material contracts: LangGraph agent, hybrid pgvector + full-text retrieval, and a code-level validation step that checks every citation against the retrieved text before the answer is shown to the user. Evaluated with RAGAS on a golden set of 40 hand-verified cases: 0 fabricated citations, 0.97 faithfulness on single-document questions. Deployed on Cloud Run with Langfuse tracing on every turn. 📌 Built an agentic receipt-capture and expense-splitting system: a Telegram bot sends the receipt photo to a vision model, extracts the line items, asks the user to confirm each one, and writes the result to Supabase, Google Sheets, and Gmail. 📌 Built a computational pathology pipeline predicting treatment response from H&E whole slide images using vision foundation models, AUC 0.788 on an 18-patient cohort, establishing proof of concept for a larger prospective study. 📌 Designed an LLM extraction pipeline to structure clinical information from oncology pathology reports, 0.88 accuracy across three cancer types, benchmarked against multiple open-source biomedical models. Manual data entry for cancer registries is a known bottleneck in clinical workflows; a pipeline like this has the potential to significantly reduce that burden, freeing clinical staff for higher-value tasks and improving data consistency across registries. 📌 Improved demand forecasting accuracy by 5% for manufacturing components using LightGBM, reducing exposure to contractual penalties tied to production shortfalls. How I work ✅ I start by understanding the business problem, what decision the output needs to support, what the data actually looks like, and what success means in your context. ✅ I work in structured phases: problem definition, data exploration, modeling, evaluation, and delivery. You know where we are at every stage. ✅I communicate proactively, you get updates at each milestone and relevant decisions or changes are communicated as they arise. ✅ I deliver documented, working code your team can maintain and build on. I hold a BSc in Mathematics, an MSc in Data Science, and I am currently a PhD researcher in Data Science at NOVA IMS in collaboration with the Champalimaud Foundation, one of Europe's leading biomedical research centres. My work spans insurance, telecoms, IT, and biomedical research. Different industries bring different ways of framing problems, and that range of experience shapes how I approach new ones.

Expert-Vetted on Upwork (Top 1%). Fractional CTO and AI engineer. 15+ years building search systems, ML platforms, credit engines, and recommendation systems across biotech, fintech, and martech. The implementation layer is getting cheaper every quarter. Judgment about what to build, when, and with what tradeoffs has gotten more valuable. That's where most of my time goes now. 𝗪𝗵𝗮𝘁 𝗜 𝗱𝗼 • Help you decide what's worth building • Help you scope it without painting yourself into a corner • Take the technical side off your plate for a while 𝗛𝗼𝘄 𝗜 𝘄𝗼𝗿𝗸 • I'll tell you not to build it if it's not worth building • I push back, and I expect you to push back on me • I take responsibility, and I expect the authority that goes with it 𝗦𝗲𝗹𝗲𝗰𝘁𝗲𝗱 𝘄𝗼𝗿𝗸 • Medical-legal document extraction. Owned architecture, build, and the calls about what to automate vs leave to humans. Weeks of analyst review compressed to minutes. The interesting problems weren't the AI ones. • Scientific AI platform for biotech R&D. Took a working PoC to a system the team actually relies on. Parser reliability, confidence calibration, CI/CD benchmarking. • AI-assisted event marketplace. Backend architecture, payments, AI planning assistant. • Domain-adapted speech recognition for spacecraft operations. Real-time voice-loop capture, indexed searchable transcripts, deployed across multiple missions. • Personalized footwear production pipeline. Parametric CAD + automated manufacturing instructions. 2 days → 30 minutes per pair. Made the unit economics work. 𝗘𝗻𝗴𝗮𝗴𝗲𝗺𝗲𝗻𝘁𝘀 • Initial Fit Call — free. 30-min call + follow-up memo with 2–3 ranked next steps. • AI Product Audit & Roadmap — from $7k, ~2 weeks. Pressure-test scope, architecture, stack, failure modes, team shape. • Fractional CTO — custom, usually 10–20 hrs/week for 3–6 months. I own architecture, technical direction, and execution quality. Stack often includes: Python/FastAPI, TypeScript/React/Next, Postgres, GCP/AWS, OpenAI/Anthropic, RAG, vector search.

Data Scientist from the University of Edinburgh specialized in Machine Learning, Data Analysis, Business Intelligence and Statistics. Five years of professional experience working in Data, Analytics and Marketing areas. I have developed consulting projects with Lloyds Bank to predict network attacks using Machine Learning techniques such as Feature Engineering; and a project with Space Intelligence to fill the gaps in satellite images using XGBoost models. Not to mention some other activities to predict and classify using supervised and unsupervised learning with Python and R. I am currently developing an End-to-end Data Science Project using optimization with Python to increase the conversion rate, and optimize their processes. I have experience using the following software: AWS. Quick Sight, S3, Transcribe, Comprehend. • Python. TensorFlow, Scikit-learn, Pandas, Seaborn, Matplotlib, Numpy, XGBoost, Keras, PySpark, Scipy. (Jupyter, VSC, Colab). • R. JAGS, INLA. • Google Cloud Platform. BigQuery, Looker, Dataproc, Pub/Sub, Dataflow, Data Catalog, Dataprep. • SAS, Minitab, SQL, Salesforce, Bizagi, Neo4j, Cypher, Tableau and Power BI. To finish I would also like to highlight that my constant work and willingness to become a better professional helped me to achieve a fully funded scholarship from the University of Edinburgh and to be a winner of the AWS Hackathon BBVA in 2020, being the first generation to achieve it.

I build AI agents and workflows faster ⚡ Want to know how? Let's chat. I’m a Data Scientist and AI Engineer with over 4 years of hands-on experience building AI systems, automation workflows, and data-driven applications for clients across multiple industries. Recently, I’ve been helping founders and companies deploy AI agents at record speed while reducing costs using the Hermes Agent, OpenClaw, Grok Bot, and other harnesses. 🌍 I speak: English, French, Spanish, and Portuguese. 🤖 I use: Cursor, Claude, Hermes Agent. ⭐ Highlights ↳ Created, designed, and deployed an agent that currently generates $45k+ MRR. ↳ Helped set up second brains with Gbrain, Obsidian, and custom solutions for personal and enterprise usage. ↳ Built production-ready Retrieval-Augmented Generation (RAG) systems using modern AI Engineering frameworks and vector databases. ↳ Built custom AI-driven automation systems for content creation, including avatar videos, infographics, and social media workflows. ↳ Trained 5+ founders to use the Hermes Agent to optimize their workloads. ↳ Developed 5+ MVPs that successfully secured funding. ↳ Built and currently maintain scalable data pipelines. ↳ Built MCPs for Claude MCP Marketplace and Cursor. ↳ Built AI agents that run on Slack, Telegram, Discord, and other gateways, including custom web applications. ↳ Built tailored agent skills and workflows for clients across multiple industries, using prompt engineering best practices. ↳ Founded a European-based agency on Upwork specialised in Harness Engineering and AI Engineering. 🧠 Skills & Tools ↳ AI Agents & RAG Systems: LangGraph, LangChain, Google ADK, ChromaDB, Vertex AI, OpenAI, Gemini. ↳ Harnesses: Cursor, Claude, DeepSeek, Hermes Agent, OpenClaw, Grok Bot ↳ Automation & Scraping: Playwright, BeautifulSoup, Apify, Camoufox. ↳ Data Science & ML: Pandas, NumPy, Scikit-Learn, PyCaret, TensorFlow, MMM (Marketing Mix Model), NLP, Optuna. ↳ Dashboards & Visualization: Plotly Dash, Data Studio, React. ↳ Web Development: FastAPI, Flask, CSS, HTML, Jinja2, Vibe-coding. 🗓️ Actively building agents for: ↳ Customer support ↳ Project management ↳ Social media posting ↳ Accounting ↳ Chat platforms like ChatGPT ↳ Slack, Telegram, Discord, WhatsApp If you need an AI agent that actually works, an automation that doesn't constantly break, or a reliable data platform, send me a message and we'll help you design the right solution.

I build AI applications that work in production. LLM agents, RAG systems, and predictive models, designed and deployed for real users. I have two years of professional experience shipping AI features in a product role, where I worked with engineering, data, and UX teams to launch LLM and RAG applications. Since then, I have moved into AI engineering and data science, building and deploying AI systems. What I can build for you: - LLM applications and AI agents using LangChain, LangGraph. - RAG pipelines and knowledge retrieval systems over your documents, databases, or internal knowledge bases. - Predictive models for classification, forecasting, and recommendations using scikit-learn, PyTorch, and explainability tools. I have trained ML models and shipped multiple production agent systems with tool use, multi-agent routing, and memory. All project code is public on GitHub. Stack: Python, LangChain, LangGraph, FastAPI, PyTorch, AWS, Docker, Kubernetes, PostgreSQL, Neo4j, Scikit-Learn, MLflow. If you have an AI project that needs to go from idea to working system, send me a message and I will put together a plan with a timeline and deliverables.