Machine Learning Research Intern, Audio

Bland•San Francisco, CA

1d•Onsite

About The Position

As a Research Intern at Bland, you will own a focused research project across our voice stack: speech-to-text, large language models, neural audio codecs, or text-to-speech. You will work alongside our research team on the same problems they are working on, not on a side track built to keep interns busy. We scope internships around a single meaningful question that can be answered in the time you have. The goal is a result worth shipping, publishing, or both. Interns here regularly see their work reach production systems handling millions of calls.

Requirements

Currently pursuing a MS or PhD in ML, CS, EE, or a related field, or equivalent research experience.
Comfortable reading a paper and reimplementing it without hand-holding.
Experience with self-supervised, generative, or multimodal modeling.
Hands-on work with speech or audio models, whether TTS, ASR, codecs, or audio representation learning.
Strong intuition for audio quality and what makes synthetic speech sound wrong.
Fluent in PyTorch and comfortable in a real codebase.
Able to run your own experiments on GPU clusters without waiting to be unblocked.
You identify the single experiment that validates an idea in days, not months.
You measure everything and let data drive decisions.
You are honest about negative results, because they are how we narrow the search.
You are obsessed with making voice agents sound truly human.
You use AI tools aggressively to amplify your own impact.

Nice To Haves

Prior publications or open source contributions in speech or language AI are a strong signal, though not required.

Responsibilities

Own a research question end to end
Take one well-scoped problem from literature review through implementation, experimentation, and results.
Design ablations that isolate what actually caused an improvement.
Present your findings to the research team and defend the methodology.
Work on real systems
Train and evaluate models on large-scale, real-world telephony audio, including the accents, noise, and artifacts that make production speech hard.
Use our distributed GPU infrastructure rather than toy-scale setups.
Where the result warrants it, work with engineers to move it toward production.
Depending on your background and interests, your project may focus on: Expressive and controllable text-to-speech, including prosody and emotion modeling; Neural audio codecs and discrete or continuous speech representations; ASR robustness for telephony, accents, and code switching; Real-time and streaming inference under latency constraints; Full-duplex conversation and turn-taking dynamics