Yassir Laaouach
making things. thinking out loud.
Currently working on agent safety after spending time building agents myself. My path started with LLMs and OCR for low-resource languages at Sussex’s CAL lab, then moved into time-series modelling and uncertainty quantification, before I worked on the systems side writing software and building AWS infrastructure for a young hedge fund, where the focus was on latency and scale. Uncertainty keeps drawing me back, each time with different tools and a new set of questions; right now I’m looking at how trustworthiness changes as agents scale and coordinate, and how to push the safety–usefulness Pareto frontier outward.