Luc Angelelli

I developed and successfully tested an algorithm on a simulator that allows a computing system, under certain assumptions, to operate at 50% of its power while maintaining over 90% quality of service.

I was invited to present my work in May 2024 in San Francisco at an international conference (International Parallel & Distributed Processing Symposium (IPDPS24)). This work was the subject of a scientific publication in December 2024 (https://link.springer.com/chapter/10.1007/978-3-031-74430-3_10).

My algorithm is written in C++ for the Batsim simulator; tests, data processing, and performance evaluation were carried out using Python, R, and shell. For part of my work, I deployed my simulator on the high-performance computing test platform Grid5000.

I taught first-year computer science undergraduates. My course covered the history of computer science, its foundations, and recent developments, as well as the basic concepts of Python and algorithms. I designed personalized learning exercises in response to specific difficulties encountered by students. This approach allowed me to adapt pedagogy to students' needs, facilitating understanding of key concepts. My goal was to make computer science accessible and engaging, while establishing a dynamic and interactive learning environment. This experience reinforced my interest in teaching and the transmission of technical knowledge.

An Edge platform is a concept close to the Cloud, but physically near users. It consists of connected objects, often with low computing power (connected doorbell, refrigerator, IoT, etc.).

The goal of this research was to evaluate whether it is practical to use the Edge rather than the Cloud for simple and frequent computations like detecting movement, for example. I created a Python algorithm that minimizes both data transmission between Edge nodes and computation time.

This work is part of a publication for the International Symposium Cluster, Cloud and Internet Computing (CCGRID23), May 2023.