Model Compression

Project description

Do you want to shape the future of video collaboration and redefine what it means to communicate, connect, and create? We are at the forefront of audio AI innovation, combining creative design, innovative science, and cutting-edge technology to deliver exceptional communication experiences. Imagine your work being at the heart of conference

rooms across the globe, used by thousands every day—including over 350 companies in the Fortune 500. We're on a mission: to make virtual communication so seamless and powerful that it's better than being there in person. As part of our growing Real-time Audio AI Team, you'll play a pivotal role in shaping leadership in AI for best-in-class, real-time audio experiences. By leveraging state-of-the-art multi-modal AI technologies, your impact will resonate on a global scale.

Responsibilities

• As a Model Compression & Optimization Engineer, you'll help build transformative AI-driven audio solutions. Your contributions will directly drive innovation in areas like model compression, efficient audio neural networks architectures, edge AI deployment. Role responsibilities will be tailored to leverage candidate's expertise
• research-oriented candidates will focus more on novel technique development and experimental work, while industry-experienced candidates will emphasize hands-on production implementation and optimization. Here's how you'll make an impact:
• Design and implement compression strategies for audio/causal models (training-time and post-training).
• Bridge Python prototyping with C++ production teams.
• Hardware-aware optimization for target platforms.
• Research cutting-edge compression techniques.
• Guide C++ team on compression method implementation.

Skills

Must have

• Strong Python proficiency.
• Solid deep neural networks knowledge (architectures, training principles).
• Experience with deep learning frameworks (PyTorch preferred).
• Understanding of model compression techniques: quantization, pruning, knowledge distillation, etc.
• Familiarity with diverse layer types (conv, recurrent, transformers).
• Hardware constraint awareness for edge deployment

Nice to have

• Preferred Experience
• General C++ knowledge to support embedded deployment of ML solutions.
• Research experience or strong academic recommendations.
• Neural Architecture Search experience.
• Audio processing or causal model experience.
• Leadership/mentoring capabilities (for senior candidates).
• Soft Skills
• Passionate about team collaboration, valuing collective success over individual output.
• Excellent attention to detail, alongside pragmatic problem-solving.
• Self-starter with the ability to conduct independent research and refine priorities
• efficiently.
• Open to feedback, with a hunger to grow professionally and personally.
• Strong communication and time-management skills to deliver impact in a fast-paced
• environment.