Research Scientist Intern, Multimodal LLM and Speech-to-Speech Translation (PhD)

What you'll do

• Conduct original research and develop models for multimodal understanding, speech modeling, or cross-language speech generation.
• Design experiments and evaluate model quality, robustness, latency, and computational efficiency.
• Collaborate across research and engineering teams and communicate results through presentations, reports, and publications.

What they ask for

• Currently pursuing or have a PhD in a listed or related STEM field.
• 3+ years programming in Python or a similar language and 3+ years using machine learning frameworks such as PyTorch or TensorFlow.
• 2+ years developing computational models in speech processing, natural language processing, multimodal learning, or audio-visual learning.
• Must obtain and maintain work authorization in the country of employment; intend to return to a degree program after the internship.

Nice to have

• Experience pretraining or fine-tuning large models with distributed training and large-scale datasets.
• Research contributions demonstrated through publications at venues such as NeurIPS, ICML, ACL, or Interspeech.

Pay

$7,650/month to $12,667/month (base salary)

Tools

Python, PyTorch, TensorFlow

AI focus

Research and development of multimodal LLMs, speech foundation models, and speech-to-speech translation, including multimodal alignment, instruction tuning, streaming inference, and expressive multilingual speech.