Youssef Miled
Hi, I'm Youssef, a Master's student in Operations Research at UC Berkeley, with a deep passion for mathematics and computer science. I was previously a research intern in the SprintML lab at CISPA Helmholtz Center for Information Security, where I worked on unlearning and privacy auditing for language models, and on memorization in generative image models. Throughout my academic and research experiences, I've cultivated skills in pure mathematics, algorithms, programming, and machine learning.
These days I am mostly interested in reinforcement learning environments, formal verification, evaluation of machine learning systems, and ASI (Artificial Super Intelligence).
Before Berkeley I studied at Centrale Lyon, and in the MP2I/MPI preparatory classes at Lycée Champollion (a program designed to prepare students for the entrance to the top French 'Grandes Ecoles') where I developed a strong foundation in pure mathematics, computer science, and physics.
Interests
formal verification and proof automation · evaluation of agentic systems · RL environments · privacy, unlearning, and memorization in LLMs · alignment and safety
Selected coursework at Berkeley
Machine Learning and Data Analytics I & II (transformers, diffusion, SFT / LoRA, RLHF / DPO, RL) · Mathematical Programming I & II · Applied Stochastic Processes · Stochastic Optimization for Machine Learning
Tools
Python · OCaml · C · PyTorch · TypeScript · React · Next.js · FastAPI · Docker · Lean 4