Principal Applied Scientist, Real-Time Conversational AI , AGI

What you'll do

• Build and scale multimodal speech and audio foundation models, including architectures and training methods suited to real-time production.
• Develop reward models, reinforcement-learning methods, post-training pipelines, and evaluations to improve conversational behavior.
• Set research direction and work with inference engineers to deliver multimodal capabilities within real-time latency constraints.

What they ask for

• PhD in Computer Science, Machine Learning, Electrical Engineering, Speech/Audio Processing, or equivalent quantitative field.
• 10+ years of applied research experience (industry or combined academic + industry).
• Hands-on experience training large-scale foundation models and experience with multimodal architectures across speech, text, and audio.
• Strong understanding of transformer architectures for speech/audio; top-tier publication record; demonstrated Principal/Staff+ scientific leadership.

Nice to have

• Hands-on experience building real-time AI systems and applying post-training methods such as reinforcement learning or reward modeling.
• Experience with distributed training at scale, speech/audio generative models, or shipping research models to production.

Pay

228,700.00 - 309,400.00 USD annually (base salary); 198,900.00 - 269,000.00 USD annually (base salary); 198,900.00 - 269,000.00 USD annually (base salary)

Tools

foundation models, transformer architectures, speech, audio, reinforcement learning, reward modeling, RLHF/RLAIF, SFT

AI focus

Researches and develops real-time multimodal conversational AI, including speech/audio foundation models, reward modeling, reinforcement learning, alignment, and real-time perception and generation.