I started out as an electrical engineer, building systems that had
to work in the real world, and that mindset has shaped everything
since. My first real test was at VOGA, where I went from intern to
team lead in under two years. I led seven engineers through the
company's acquisition by BTG Pactual, Latin America's largest
investment bank, and designed the real-time portfolio platform that
tracked over USD 300M in client investments. The more I built, the
more I wanted to study AI itself, so I earned a Japanese Government
(MEXT) scholarship and moved to Tokyo. At the University of Tokyo, I
worked side by side with Nikkei, owner of the Financial Times, on
how millions of people really read the news. I built news
recommenders that treat the stories readers avoid as implicit
negative feedback to counter popularity bias, multi-agent LLM
pipelines that use retrieval-augmented generation to extract reader
behaviour patterns, and NewsReX, an open-source JAX and PyTorch
framework for reproducible news recommendation research. With
Toyota, I turned GPS vehicle telemetry into trajectories and
user–place graphs, then trained graph neural networks for
next-destination prediction, benchmarking them against a range of
baselines on Toyota's proprietary driving data. Most recently, I
completed a Master of Research in AI and Machine Learning at
Imperial College London, where I studied why contrastive learning
and mixture-of-experts models collapse when data is scarce.