Senior Real-Time Multimodal AI Scientist

Amazon Inc.; Seattle, Northern (WA, KY); Hybrid; USD 192,000 - 260,000; Full time

Job summary

Amazon is seeking a Senior Applied Scientist to advance real-time multimodal conversational AI, spanning foundation models for speech and audio and post-training systems shaping natural conversations.

You will own a significant research area across pre-training, architecture design, and deployment, working with inference engineers to ensure models are production-ready with real-time performance.

Qualifications

• 5+ years building ML models for business apps
• PhD or Masters with 6+ years of applied research
• Proficient in Java, C++, Python or equivalent
• Hands-on training of large-scale foundation models
• Experience with multimodal architectures across speech/text/audio
• Strong grasp of transformer architectures for speech/audio
• Publication record at top venues
• Ownership of research areas and cross-team collaboration

Responsibilities

• Drive scaling of real-time multimodal foundation models
• Design architectures mindful of hardware and latency constraints
• Build reward models and RL methods for speech systems
• Develop post-training pipelines from SFT to RL alignment
• Collaborate with inference engineers for production deployment

Skills

Java

C++

Python

Multimodal architectures

Transformer models

Education

PhD or Masters + 6+ years

Tools

TensorFlow

PyTorch

JAX

Job description

Amazon is seeking a Senior Applied Scientist to advance real-time multimodal conversational AI, spanning foundation models for speech and audio and post-training systems shaping natural conversations.

You will own a significant research area across pre-training, architecture design, and deployment, working with inference engineers to ensure models are production-ready with real-time performance.