Sr. Principal Software Scientist

Who is Cerence AI? 

Cerence AI is the global leader in AI for transportation, specialized in building AI and voice-powered companions for cars, two-wheelers, and more that enable people to focus on what matters most. With over 500 million cars shipped with Cerence AI's technology, we partner with leading automakers (such as Volkswagen, Mercedes, Audi, Toyota and many more), mobility providers, and technology companies to power intuitive, integrated experiences that create safer, more connected, and more enjoyable journeys for drivers and passengers alike. 

Our Driving Force  

Our team is dedicated to pushing the boundaries of AI innovation, working around the globe with headquarters in Burlington, Massachusetts, USA and 16 other offices across Europe, Asia, and North America. We bring together diverse backgrounds, and varied skill sets with the shared goal of advancing the next generation of transportation user experiences. Our culture is customer-centric, collaborative, fast-paced, and fun, with continuous opportunities for learning and development to support your career growth. 

Interested in having a significant impact in a dynamic industry with a high-performing global team? We’re looking for an exceptional Senior Principal AI Scientist in Generative AI who is ready to drive the future of mobility with us! 

What You Will Work On 

• Design and train large‑scale transformer and hybrid foundation models 

• Own model architecture choices across text, multimodal, and emerging paradigms 

• Diagnose and resolve training instabilities at scale 

• Navigate scaling tradeoffs across data, compute, and architecture 

• Define the technical direction for next‑generation models 

Core Responsibilities 

Deep Learning & Transformer Foundations 

• Apply strong fundamentals in deep learning and representation learning 

• Design and modify transformer architectures, including: 

• Attention variants 

• RoPE, ALiBi 

• Grouped Query Attention (GQA) 

• Mixture‑of‑Experts (MoE) 

• Build models from first principles, not just adapt pre‑existing codebases 

Optimisation Dynamics & Training Stability 

• Own optimizer and scheduler choices, including: 

• AdamW 

• Lion 

• Adafactor 

• Learning‑rate and warmup schedulers 

• Understand and debug: 

• Optimizer instability 

• Gradient pathologies 

• Divergence at large sc