About

Alícia Oliveira is a Machine Learning Engineer at INESC TEC, where she contributes to the EPICURE EuroHPC JU project, and currently also provides user support at the Deucalion Supercomputer in Portugal. Her work lies at the intersection of large language models (LLMs), machine learning, and high-performance computing (HPC), with a particular focus on the efficient deployment, optimization, and scaling of AI workloads in advanced computing environments. Her research interests include LLM systems, inference and serving frameworks, distributed execution, performance analysis, and infrastructure support for AI at scale. She is currently pursuing a PhD in Informatics at the University of Minho, with a thesis entitled “A systems perspective for LLM optimizations”. She previously completed a Master’s degree in Medical Informatics with the dissertation titled “Federated Learning-based Artifact Correction in Brain MRI scans in HPC environments” and holds a degree in Biomedical Engineering.  Throughout her academic and professional trajectory, she has developed solid expertise in LLM and deep learning systems, spanning model development, training, and inference workflows, performance tuning, monitoring, experiment tracking, and scalable deployment. Her technical experience includes working with LLM serving and inference systems such as vLLM and llama.cpp, as well as supporting and optimizing multi-node and heterogeneous CPU/GPU environments for AI workloads. Her background also includes experience with PyTorch, TensorFlow, federated learning, MLflow, TensorBoard, and HPC-oriented AI workflows, with a growing specialization in the systems-level challenges of running and optimizing LLM-based applications across large-scale computing infrastructures. She has been actively involved in technical support, user training, and dissemination activities related to advanced computing, while also contributing to lectures, outreach initiatives, hackathons, and international events in the fields of AI and HPC, including the EuroHPC Summit, SC25, and the ACM Europe Summer School on HPC Computer Architectures for AI. Her publication record includes research in neuroimaging and artifact correction, with a conference paper presented at IEEE CBMS 2026, reflecting a broader interest in applying machine learning to scientific and real-world challenges. Her current work is increasingly focused on LLM efficiency, systems optimization, and AI at scale.