After a successful internship in the Barcelona office, I have been invited to do an internship at the headquartersof the company in Silicon Valley. Mainly doing research on ways to compress the KV Cache of Large LanguageModels, using both quantization and eviction based approaches. This is done in order to decrease thecomputational and energetic cost of them during inference.

Providing assistance on the Mathematical Optimization subject of the degree in Data Science and Engineering

Esperanto Technologies is a tech startup which has its headquarters located in Mountain View (California),focused on energy efficient hardware, HPC and AI. I worked on implementing diffusion models in their pipeline,gaining deep insight on how these models work from an algorithmic and theoretical standpoint. My work wasintegrated as a core element in the software stack of the company.

Did research on applications of Deep Learning for Survival Analysis problems, focused on a specific clinical case. I familiarized myself with different neural network architectures like Transformers and RNNs, as well as some concepts in Survival Analysis like Kaplan Meier Curves or Cox Proportional Hazards Models