Talk, IA Sob Controle,
In this talk (in Portuguese), I walk through the hurdles and bottlenecks involved in developing GigaVerbo—the largest Portuguese text dataset at the time of publication—and the Tucano Series.
Talk, IA TRENDCAST,
In this talk (in Portuguese), I discuss the ethical, political, and technical challenges of building AI in and for Brazil—ranging from value alignment and digital sovereignty to the creation of Tucano and GigaVerbo—while reflecting on the broader landscape of Portuguese‑language AI and the need for more open, locally grounded innovation.
Talk, C4AITech hour (USP),
In this talk (in English), I walk through the development of GigaVerbo—the largest Portuguese text dataset at the time of publication—and the Tucano Series, while also situating both within the recent historical evolution of Portuguese‑language LLMs.