10/13/2026 updated
40 % partially available
AI Engineer · AI Agents & Document Intelligence
Porto, Portugal
Only remote
M.Sc. Software Engineering, FEUP (2024–2026); B.Sc. Informatics and Computing Eng., ISEP (2021–2024)About me
I build AI agents and document extraction systems that run in production, for regulated companies where accuracy, traceability and data protection matter.
AI AgentsPython (Programming Language)Multi-Agent Systems
I'm an AI Engineer at DareData in Porto, Portugal. For more than 3.5 years I've built production AI agents for regulated clients in insurance and finance.
Recent work:
- An agentic document tool used by about 90 people at msg insur:it.
- Configure:it, a multi-agent module.
- FreeFloat, a browser agent for Euronext. It covers 614 companies and saves about 120 person-days a year.
I also maintain open-source tools. complydoc (PyPI, MIT, sole maintainer) analyses documents before they reach an AI pipeline. It covers ingestion observability, comparing document loaders, and CI gates for personal data and prompt injection. article-explainer has 190 GitHub stars and was featured by LangChain.
I hold an M.Sc. in Software Engineering from FEUP (18/20, thesis 19/20), and received the SEI 2025 Best Paper award.
I help with three kinds of work: designing and building AI agents, turning unstructured documents into reliable structured data, and taking an AI prototype to a dependable production system. Each project starts with a clear scope and agreed success criteria, and ends with documentation and a handover your team can own.
I work in English, Portuguese and Spanish, and I'm especially keen to work with teams in France, Germany and Benelux.
Experience
Professional experience0 - 5 years
Freelancer since2026
Languages
EnglishFluentPortugueseNative speakerSpanishFluent
Project history
AI consultant for Euronext, designing, developing and deploying cloud-native AI solutions on AWS and Azure that automate operational workflows.
Main developer of FreeFloat, a Dataiku-integrated browser agent (Strands, AWS) automating free float review across 614 companies, targeting ~120 person-days saved annually across four review cycles.
Built the company's official Document Intelligence tool on an Agentic RAG architecture, serving ~90 users company-wide since launch.
Designed and developed Configure:it, a multi-agent Product Machine module converting documents into product configurations and enabling natural language querying.
Developed my Master's thesis in-house, the Product Validation Accelerator Engine, which combines LLM agents with combinatorial testing, boundary value analysis and equivalence partitioning to turn a plain-language request into a complete test suite, cutting insurance product validation from ~6 months of manual client work to a handful of automated requests.
Supported a colleague's thesis on LLM-based extraction of insurance product characteristics from regulatory documents, leading to two co-authored papers (see Publications).
Full-stack engineer on an enterprise-grade insurance web application for a North American client, in an international, cross-functional team (Java Spring, Backbone.js, PostgreSQL); contributed to a successful go-live with a very low defect rate.
As an intern, built an end-to-end insurance product proof of concept, implementing workflows with Camunda (BPMN/DMN) integrated with Java services.