Hi, I’m Gabriel!

This is my research blog, where I document my PhD work in Computer Science. Here I share research notes, technical write-ups, experimental insights, and reflections that emerge from my doctoral studies.

My current research focuses on weak supervision for semantic segmentation, with an emphasis on learning reliable dense representations under limited, noisy, or indirect supervision. I am particularly interested in understanding how different weak signals, such as partial annotations, proxies, or priors, can be leveraged to scale segmentation systems beyond fully supervised settings. This work builds on my prior experience with seismic data analysis and seismic facies segmentation, a domain I expect to continue exploring as both a motivating application and a source of challenging real-world constraints.

This space is intentionally focused on my academic work. Content is written primarily for researchers, students, and practitioners interested in modern machine learning systems and representation learning.