Johns Hopkins Sophomore Alessa Carbo Publishes Landmark Sign Language AI Paper

Context Alessa Carbo, a computer science student at Johns Hopkins, arrived at university already adept in programming and machine learning. She developed these skills independently in Cabo San Lucas, Mexico, a town without a public library or high school computer science courses. Carbo stated, "I had internet access and an intense curiosity about the world," driving her self-education.

Her exploration led her to Johns Hopkins' Center for Language and Speech Processing (CLSP) seminars and the Artificial Intelligence Society. These experiences introduced her to sign language processing, an area she identified as unique and underexplored within natural language processing. This field combines computer vision, linguistics, and natural language processing.

Key Facts In summer 2024, Carbo joined the Frederick Jelinek Memorial Summer Workshop, collaborating with a Czech team. She helped develop a custom vision-language AI model, a system that processes visual information alongside textual data. This model translates sign language video directly into English text.

Her work culminated in a first-authored paper accepted by the 2025 Conference on Empirical Methods in Natural Language Processing (EMNLP). EMNLP is a leading international scientific conference on natural language processing. Eric Nalisnick, her mentor and an assistant professor of computer science, noted this achievement. He stated that securing a first-author paper at EMNLP is a significant accomplishment for a PhD student, making it especially notable for a sophomore carrying a full academic course load.

What It Means Carbo's publication highlights the potential for early and significant contributions in advanced AI research. Her journey from self-taught programmer to published AI researcher underscores the impact of individual initiative and institutional support. She has since shifted her primary research focus to AI safety and joined the Machine Learning Alignment and Theory Scholars fellowship program. She continues to collaborate on projects, including another first-authored paper slated for the 14th International Conference on Learning Representations. Watch for her continued contributions across diverse AI domains, particularly in AI safety and interpretability.