Improving the robustness of Reinforcement Learning for real-world applications
My name is Juan José (Juanjo) and I'm a PhD Candidate in the Engineering Systems Laboratory at MIT. I work on improving the robustness of Deep Reinforcement Learning-based autonomous systems in real-world environments, both from domain-agnostic and domain-specific perspectives. I focus on studying and solving the many challenges this entails: generalizability, non-stationarity, multi-objectiveness, sample efficiency, stability, etc. In addition to conceptualizing new ways of designing, implementing, and operating RL in the real world, I work in the application of RL for dynamic resource management in satellite communications, molecular optimization, and traffic signal control.
In 2017, under the CFIS program, I received two BSc degrees in Telecommunications Engineering and Industrial Engineering from UPC, in Spain. I carried out my undergraduate Thesis as a visiting student in the System Architecture Lab at MIT; I designed a large-scale Internet of Things network by optimizing its architecture leveraging systems engineering. I previously worked at Arcvi and at Barcelona Supercomputing Center. In 2019, I was awarded a La Caixa fellowship with full funding for my PhD studies. In 2020, I received my Master's degree in Aerospace Engineering from MIT. My Master's Thesis focused on the study of autonomous resource management systems for communication satellites and the use of various AI algorithms to support their operation. Later that year, I interned at Novartis AI Innovation Lab. In 2021, I interned at Google Brain, where I remained as a Student Researcher throughout the rest of the year and part of 2022.
Deep Reinforcement Learning
Applied Machine Learning