Speech Research Intern – 2

This is a fully remote position, open to applicants in United States.

📋 Description

• Create and assess speech-first models, with an emphasis on Spoken Language Models (SLMs)

• Transition concepts from prototypes to functional demonstrations

• Collaborate with scientists and engineers to achieve measurable outcomes

• Build comprehensive speech dialogue systems and speech-aware large language models (LLMs)

• Maintain consistency between speech encoders and text backbones through lightweight adapters

• Perform reliable evaluations across tasks related to recognition, understanding, and generation

• Implement latency-aware inference to enhance real-time user interactions

⛳️ Requirements

• PhD candidate in Computer Science/Electrical Engineering (or related field) with research experience in speech, audio machine learning, or multimodal language models

• Proficiency in Python and PyTorch, along with practical experience in GPU training

• Familiarity with torchaudio or librosa

• Solid understanding of contemporary sequence models (Transformers or State Space Models) and best practices for training

• Expertise in at least one specific area: (a) discrete speech tokens or temporal compression, (b) modality alignment to LLMs using adapters, or (c) post-training or instruction tuning for speech-related tasks

• Strong experimental skills: clean coding practices, ablation studies, reproducibility, and effective reporting

🏝️ Benefits

• Competitive stipend

• Guidance from applied scientists and engineers

• Opportunities to publish and present research

• Access to state-of-the-art GPU infrastructure

• Supportive setting for rapid and responsible experimentation

• Flexible options for location and scheduling