The Team and Role:
As an Audio ML Engineer on the Logitech Hardware Audio ML and DSP Product team, you will be instrumental in developing Embedded Audio ML models that create innovative audio experiences for our customers [ e.g. Speech/Audio enhancement] . This role offers a significant opportunity to contribute directly to the audio products we develop.
The Audio ML Data Engineer's key responsibilities include:
- Develop production-ready Audio ML models, leveraging multi-sensor data from the product.
- Ensure these models are deployed for efficient inference on resource-constrained platforms by employing optimization techniques, including Post-Training Quantization (PTQ), Quantization-Aware Training (QAT), pruning etc.
Your Contribution:
Be Yourself. Be Open. Stay Hungry and Humble. Collaborate. Challenge. Decide and just Do. Share our passion for Equality and the Environment. These are the behaviors and values you’ll need for success at Logitech. In this role you will:
- Develop and implement highly optimized Audio ML models for efficient deployment on resource-constrained embedded platforms (e.g., ARM, Tensilica DSP, RISC-V, NPUs).
- Utilize techniques like quantization [PTQ and QAT], and pruning to ensure effective on-device inference.
- Architect, optimize, and improve algorithm performance in complex real-world audio environments.
- Propose and implement novel solutions to challenging technical problems.
- Collaborate with various product teams to guarantee a premium and seamless customer audio experience.
Key Qualifications:
For consideration, you must bring the following minimum skills and experiences to our team:
- Audio ML Expertise (3+ Years): Hands-on experience across the entire Audio ML lifecycle, including model training, tuning, quantization (PTQ and QAT), and deployment to production.
- ML Framework Proficiency: Advanced skills in ML frameworks (e.g., TensorFlow, Keras,PyTorch) with a history of successfully shipping Production ready Audio ML models.
- Embedded Optimization: Demonstrated success in optimizing model inference performance specifically for resource-constrained embedded systems.
- Strong Programming & Best Practices: Excellent programming skills in Python and C, coupled with experience in code optimization and adherence to rigorous software best practices.
- Audio Data Augmentation: Experience with audio data augmentation techniques, including the ability to design, implement, and evaluate custom augmentation pipelines.
- ML Stack Debugging: Proficiency in Linux-based compute environments and experience debugging common ML training stack issues (e.g., OOM issues, CUDA errors, library conflicts).
Preferred Qualifications:
- Audio Quality Assessment: Proven experience in designing and executing both subjective and objective audio quality evaluation protocols, including familiarity with industry-standard audio measurement metrics.
- Audio Artifact Resolution: Demonstrated track record of effectively identifying and resolving audio artifacts within ML audio chains..
- Technical Leadership & Communication: Excellent communication, documentation, and leadership abilities, particularly in cross-functional technical environments.
- Initiative & Execution: Highly driven individual with a demonstrated ability to deliver results and lead technically, both independently and as a contributing team member.
Education:
- Bachelor’s or Master’s degree in Electrical Engineering, Computer Science, or a closely related field.
- Equivalent practical experience is considered; advanced degrees or continuing education in audio ML are highly valued.
