Machine Learning Intern

About this role

Lead an independent machine learning research project in speech and audio at Bland, working on problems like speech recognition, text-to-speech, or neural codecs using real production telephony data and distributed GPU infrastructure. Your research will directly impact millions of calls and potentially ship to production or be published.

What you'll do

• Own a research question from literature review through implementation, experimentation, and presentation to the team
• Train and evaluate models on large-scale, real-world telephony audio with production noise and accents
• Design ablation studies to isolate the causes of improvements and validate hypotheses
• Work with distributed GPU infrastructure to experiment independently without waiting for approvals
• Collaborate with engineers to move promising results toward production systems where appropriate
• Choose research direction within audio/speech domains based on interests and background

What they're looking for

• PyTorch and deep learning implementation
• Speech or audio model development (TTS, ASR, codecs, or representation learning)
• Self-supervised or generative model experience
• GPU cluster and distributed computing workflows
• Research methodology and experimental design
• Paper reading and algorithm reimplementation
• Audio quality evaluation and intuition
• Scientific communication and results presentation

Benefits

• Competitive intern compensation
• Mentorship from frontier voice AI researchers
• Access to all necessary tools and infrastructure
• Office in Levi's Plaza, San Francisco with rooftop views
• Strong potential for return offer conversion
• Work on problems with real-world impact at scale