I’m currently pursuing a career as a research engineer in AI alignment and mechanistic interpretability. Most recently, during an intensive 12-week retreat at Recurse Center, I completed the materials of the ARENA course on alignment engineering and subsequently led a new cohort of 10-15 students through the materials. I also lead a research engineering study group focused on paper reproductions.
My academic background is in deep learning, signal processing, probability theory and linear algebra. In my undergraduate thesis, conducted during an internship at Telefónica Research, I explored the use of recurrent neural networks for speech detection.
I am driven, on one hand, by the prospect of studying near-intelligent systems in hopes of understanding how they acquire and represent knowledge, which is immensely appealing to me as an avid technologist with strong cognitive science inclinations. Another major motivator is my belief that technologies as complex and capable as advanced AI are unlikely to be safe and free of major risks unless we work hard to ensure that they are. I believe that developing a mechanistic understanding of deep learning systems is an essential step towards making them understandable, steerable and safe.
Previously, I spent six and a half years at BMAT Music Innovators, first as a Barcelona-based software engineer focused on music recommendation and music rights data pipelines, and later launching the company’s US team as a direct report to the CEO. For five years, as BMAT’s product and business development lead, I owned the relationships with major labels and publishers, music platforms, rights organizations and key competitors. I led mission-critical initiatives with clients including SoundCloud, ByteDance/TikTok and Warner Music, from discovery and prototyping, through commercial and legal negotiation, to agile product execution and ongoing customer success.
A major through line that connects most of my work is the exploration of the way technology mediates human cognition, curiosity and creativity. This has inspired a number of creative and experimental projects, exploring topics such as AI-enhanced walking meditations and knowledge management; high-tech and low-tech approaches to music discovery; data-driven storytelling around political polarization on Twitter; and experimental musical instruments driven by gesture tracking, cellular automata and tangible interfaces.
I also play music, a lifelong passion, and take pictures, a recent discovery.