Teams at Cerence Inc.

Lead research and engineering work on LLM-based architectures for dialog, search, and destination entry. Optimize LLMs via prompting and fine-tuning, integrate them with ASR, TTS, and vision systems, and collaborate cross-functionally to deliver performant, resource-constrained solutions for automotive voice assistants.

Research, design, and train large-scale transformer and hybrid foundation models. Diagnose training instabilities, optimize scaling tradeoffs, define next-gen architectures, develop loss and alignment strategies (RLHF/DPO/GRPO), and lead distributed training (FSDP, ZeRO-3, tensor/pipeline parallelism) with mixed precision and gradient checkpointing while partnering with ML systems teams.