AI Scientist @ Thales Research & Technology
Speech & Audio AI · Deep Learning · Signal Processing · Edge AI
I build intelligent audio systems that operate under real-world constraints — from research and model development to deployment on embedded and edge devices.
My current work focuses on speech enhancement, generative audio, audio foundation models, and efficient deep learning, with a particular interest in real-time and resource-constrained systems.
My research background spans deep learning optimization, signal processing, and aerospace engineering. I previously worked at Mila – Quebec AI Institute, where I developed AdaFisher, a second-order optimizer accepted at ICLR 2025.
🎧 Speech Enhancement · 🤖 Generative Audio · ⚡ Edge AI · 🧠 Deep Learning · 📡 Signal Processing
- Speech Enhancement — denoising, dereverberation, intelligibility enhancement
- Real-Time & Embedded AI — lightweight models, low-latency inference, ONNX deployment
- Generative Audio — flow matching, diffusion/flow-based models, audio foundation models
- Audio Representation Learning — spectrogram and waveform modeling
- Robust Audio AI — models operating under challenging acoustic conditions
- Deep Learning Optimization — second-order optimization and neural network dynamics
- Multimodal Learning — combining audio with other modalities