Principal Applied Scientist, Real-Time Conversational AI , AGI
Core
Drive research and development of real-time multimodal conversational AI, focusing on foundation models for speech/audio and post-training systems for natural conversational behavior.
Role type
Principal Applied Scientist (Real-Time Conversational AI)
Builds
Large-scale multimodal foundation models and post-training pipelines for real-time speech and audio generation
Domain
AI/ML, Speech & Audio Processing, Conversational AI
Deliverable
production ML models
Required skills
Foundation model scaling, multimodal model architecture design, transformer architectures, reinforcement learning, reward modeling, real-time system design, distributed training, speech/audio processing
Preferred skills
Real-time AI systems, RLHF/RLAIF, speech-to-speech generation, hardware-informed model design, shipping research to production, open-source contributions
Technologies
Transformers, Reinforcement Learning, Reward Models, Distributed Training, Speech/Audio Codecs
Responsibilities
Design and train large-scale multimodal foundation models for real-time speech and audio; Develop reward models and reinforcement learning methods for conversational behavior; Work with inference engineers to ensure models meet real-time latency budgets; Design evaluation frameworks for real-time conversation quality
Seniority
Principal, hands-on IC with strategic leadership