Speech & Audio ML Research Engineer — Real-Time Edge Deployment

Ifm Us; Paris; Sur place; EUR 70 000 - 95 000; Plein temps

Résumé du poste

Institute of Foundation Models seeks a Research Engineer specializing in speech/audio machine learning to bridge research and production, delivering scalable, low-latency AI systems.

You will optimize compute graphs, apply model compression, build reliable data pipelines for audio datasets, and implement distributed training using DeepSpeed and Horovod, with deployment in Docker and WebRTC integrations.

Qualifications

• MSc or BSc in Computer Science or Software Engineering.
• Engineering Excellence: Strong background in Systems Programming and Computer Architecture.

Responsabilités

• Compute Optimization: Optimize model compute graphs for target runtimes using frameworks such as TensorRT, Apache TVM, or OpenVINO.
• Model Compression: Perform quantization (INT8/FP8) and pruning for real-time execution on edge or cloud GPUs.
• Data Orchestration: Build reliable data pipelines for audio datasets, including automated denoising and synthetic data augmentation.
• Distributed Training: Help design efficient multi-node training scripts using frameworks such as DeepSpeed, Horovod, as well as other distributed training technologies.
• Deployment: Build and maintain high-performance demo environments on the cloud using Docker and standalone model serving frameworks. Integrate WebRTC for real-time audio-to-audio interactions.

Connaissances

Compute optimization

Model compression

Data orchestration

Distributed training

Deployment

Formation

MSc or BSc in Computer Science or Software Engineering

Outils

TensorRT

Apache TVM

OpenVINO

DeepSpeed

Horovod

Docker

WebRTC

Description du poste

Institute of Foundation Models seeks a Research Engineer specializing in speech/audio machine learning to bridge research and production, delivering scalable, low-latency AI systems.

You will optimize compute graphs, apply model compression, build reliable data pipelines for audio datasets, and implement distributed training using DeepSpeed and Horovod, with deployment in Docker and WebRTC integrations.