Senior computer scientist leading the AI and software stack at Lisbon Economics — doctoral-level machine learning, current Azure AI work on Large Language Models and Agents, and a track record of GPT product integration at Microsoft.
Principal Software Engineer at Microsoft, currently working on Large Language Models and AI Agents within Azure AI — including assessment of accuracy, cost-effectiveness, capabilities, and limitations of agent systems.
PhD in Computer Science from Imperial College London (2010), specialising in logic-based machine learning. Earlier degrees in Artificial Intelligence (MSc, Universidade NOVA de Lisboa) and Bioinformatics.
At Microsoft since 2018, prior product integrations of GPT models include the tone functionality in SwiftKey and query rewriting in the Bing search engine. Earlier Microsoft Language Development Center post-doctoral work focused on query relevance. Two US patents (2016 and 2022) covering query classification and digital content service quality. Wellcome Trust Scholarship recipient for doctoral studies; Microsoft FY16 Individual Contributor Award.
At Lisbon Economics, leads the AI and software stack used across mandates — statistical, econometric, structural-modelling, and machine-learning pipelines — ensuring every engagement runs on the current state of the art.
Concurrent roles at Microsoft and Lisbon Economics — current product work on LLMs and Agents at Azure AI, and design of the AI / quantitative stack used on every mandate.
Two US patents on query classification and content quality, plus peer-reviewed research on machine learning applied to protein–ligand interactions.
Every mandate runs on a current, AI-grade quantitative stack. The list below names the principal tools — versioned, scripted, and reproducible end-to-end.
Large Language Models and Agent evaluation (Azure AI), classical machine learning, and logic-based ML applied to evidence extraction, document review, and case-pattern analysis.
Random-coefficient discrete choice, BLP estimation in pyblp, merger simulation, pass-on and damages quantum — in Stata, R, and Python pipelines.
Structural industry models, dynamic pricing simulations, and counterfactual benchmarking, calibrated to the matter and validated against academic literature.
Send a one-paragraph brief. A senior member of the network will reply within one working day with a read on fit and approach.