Experience

CEA-Leti

• Semiconductor Manufacturing
• 700 & Above Employee

• Research Engineer - AI-based optimizations for 6G wireless networks

  • Aug 2021 - Present

  🎯 Investigating machine learning techniques for distributed network management, mobile edge computing, resources allocation for wireless networks, and edge learning.• Stochastic optimizations• Multi-agent reinforcement learning• Multi-armed bandits• Constrained (non) convex optimizations🎯 Developing a Python/Tensorflow-based system level simulation tool for distributed radio resources management and orchestration using multi-agent learning.🎯 Participating in the lab roadmap definition.🎯 Driving innovative research and contributing to the growth of the company’s patent portfolio (intellectual properties).🎯 Structuring and building technical projects.🎯 Supervising Internships, PhD theses, and Postdocs. Show less

• Ph.D. - Deep multi-agent reinforcement learning for 5G and beyond communications and networking

  • Nov 2018 - Oct 2021

  🎯 Investigated distributed learning mechanisms for distributed network management, mobile edge computing, resources allocation for wireless networks.📌 Main achievements:✔️ Designed a python- and Tensorflow-based network-level & link-level simulator (I'm particularly proud of). The simulator can instantiate network agents such as Federated Learning Agents, Reinforcement Learning Agents at different levels (base stations, user devices, central orchestrator) for network radio resource management. With this simulator, we can perform centralized and distributed optimization of e.g. user association, handover, or MEC policies. Also, we can manage mobility, predict network traffic. ✔️ Proposed novel distributed user association, beam management, and handover managing mechanisms with up to 95% of the optimal performance obtained via exhaustive search approach.✔️ Managed research project with an industrial and filed two patents.✔️ Made strong technical contributions to European Projects (CPS4EU, 5G-CONNI, ...)✔️ Authored 3 patents and,✔️ Published 6+ works in top-ranked IEEE conferences and journal as the first author.📌 Index terms: Wireless Networking; Semantic Communications; (Deep) Multi-agent Reinforcement Learning; Machine Learning; Distributed Learning; Federated Learning; Natural Language Processing; Python; Tensorflow; Keras; MatLab Programming. Show less



STMicroelectronics France

• France
• Semiconductor Manufacturing
• 700 & Above Employee

• High-speed serial links modeling (IPs SerDes, 56Gbps PAM4)

  • Feb 2018 - Aug 2018

  🎯 With the increasing demand for high data rates in communication systems, new standards are rising to offer high-speed serial links while lowering the silicon surface used. This requires re-designing and optimizing communication systems. 📌 Main achievements: ✔️ Modeled algorithms for the Clock Data Recovery (CDR) of the SerDes receiver. ✔️ Optimized algorithms to boost the performance of the previously implemented communication system simulator with up 100x gain: I transform one-hour simulation into less than one minute. ✔️ Investigated and implemented more sophisticated channel equalization techniques, which allow improving communication rate to up to 5 dB. 📌 Index Terms: Communication Systems; IP SerDes; Phase Locked Loop (PLL); MLSE, Continuous Time Linear Equalization (CTLE), Decision Feedback Equalizer (DFE), FeedForward Equalizer (FFE), Maximum Likelihood Sequence Estimation (MLSE), Combined MLSE / DFE, Viterbi algorithm applied to channel equalization, (Blind) channel estimation, MatLab, C/C++ Programming Show less



GIPSA-lab

• France
• Research Services
• 1 - 100 Employee

• Anomalies detection in electrocardiogram signals using Extended Kalman Filtering

  • May 2017 - Aug 2017

  🎯 In the medical field, the electrocardiogram (ECG) signal provides a lot of information about the condition of patients. Monitoring the ECG helps to prevent heart problems. The goal of this internship was to find a way to efficiently extract fiducial points from the ECG, using Kalman filtering methods. Once interpreted, these points provide information on the condition of the patient. 📌 Main achievements: ✔️ Modeled ECG signals analytically and implemented signal processing algorithms for ECG denoising. (McSharry et al.) ✔️ Benchmarked non-linear filters (Extended Kalman Filter and Smoother, UKF) and ECG signal tracking on Matlab. ✔️ Developed and optimized embedded C algorithms for real-time monitoring of ECG signals. ✔️Collaborated with the hardware (HW) team for practical algorithm implementation. 📌 Index Terms: Bayesian filtering, Kalman filtering, electrocardiogram signals, Matlab, and C programming. Show less



TDK InvenSense

• United States
• Semiconductor Manufacturing
• 500 - 600 Employee

• Sound classification for sleep quality analysis

  • Jun 2016 - Aug 2016

  🎯 Snoring is described as a sleep disorder. It degrades the quality of sleep and in some cases is perceived as a symptom of respiratory disorder. This internship consisted in building an annotated database for sleep monitoring using an audio stream. To do so, we tracked abnormal sound signals and assigned them an annotation according to their intensity. Eventually, this allows signal classification to estimate the patient's sleep quality. 📌 Main achievements: ✔️Built an annotated database by tracking and labeling anomalous events in audio recordings for classification purposes. The internship was done at Movea, now belonging to TDK InvenSense. Index Terms: Matlab Programming, Praat, Audio Analysis, and Annotation. Show less