Research Internship
About Gradium
Gradium is a voice AI company building the Text-to-Speech (TTS) engine for the next generation of voice agents.
Born out of Kyutai, the Paris-based open science AI lab, Gradium brings frontier speech research into production. We believe the best voice agents will win on quality, and that quality starts with the voice itself. We’re purpose-built for product engineers and AI-native companies who need voices that sound genuinely human, stay reliable in production, and don’t break on real-world content like acronyms, alphanumerics, or cross-lingual names.
Our models cover the full voice stack: text-to-speech, speech-to-text, speech-to-speech, live translation, on-device inference and voice cloning.
We are a small team, early but moving fast, with real customers and a clear thesis: voice agents are the next major frontier of applied AI, and the models underneath will matter enormously.
What you’re stepping into: small team, unreasonable ambition, high ownership, fast decisions, and very little process unless it earns its place.
About the role
We’re looking for a Research Intern to join our research team and work on real problems in voice AI, across synthesis, recognition and everything in between.
Our research team is world-class: the people behind some of the most significant recent advances in speech and audio modeling. As an intern, you’ll work side by side with them, on the same problems, with the same tools and compute. It’s a rare opportunity to learn from the best, in a setting where your work doesn’t stay in a notebook: it can end up in production, in the hands of our customers.
What You’ll Do
Define your subject: With your supervisor, shape a precise research topic that fits your profile, your interests and Gradium’s research priorities. Topics are grounded in real product challenges (quality, latency, robustness, new capabilities), and will be aligned with your school or university requirements.
Master the state of the art: Dig deep into the literature on your topic. Read the papers, understand the techniques, and reproduce key results when it helps.
Form sharp hypotheses: Critically analyze existing approaches, identify where they fall short on real-world content, and formulate clear, testable hypotheses with your supervisor.
Build and train models: Design architectures and training methods, build or curate the datasets you need, run experiments at scale, and evaluate rigorously, on benchmarks and, crucially, on messy, real customer content. Write code that meets the team’s engineering standards.
Take it toward production: Work with the research and product teams to understand what it takes for a model to hold up in production (latency, reliability, cost) and help bring your results closer to something that ships.
Share your work, when it makes sense: Gradium contributes to open source and the research community. Depending on your project, you may have the opportunity to release code or contribute to a publication, though this isn’t guaranteed for every internship.
Who You Are
Strong technical foundations: You’re a Master’s student (M2) or in the final year of an engineering school or university, specializing in applied mathematics, computer science and/or machine learning.
Solid in math and algorithms: You’re comfortable with the theory and know how to turn it into working code.
Grounded in deep learning: You understand the fundamentals of machine learning and deep learning, ideally applied to audio signal processing, speech processing, natural language processing or computer vision.
Hands-on engineer: You’re fluent in Python and have experience with PyTorch, JAX or TensorFlow. Experience with C++ or Rust is a plus.
Curious and rigorous: You care about understanding why things work, not just whether they do. You hold yourself to high scientific standards and love working as part of a team.
Ownership mindset: You take initiative, move fast, and are energized by the idea of your work making a real-world difference.
Fluent with AI agents: You build and work with AI agents to speed up both your coding and your research, and you know how to keep their output rigorous.
What We Offer
Learn from a world-class research team, with close supervision and the chance to work on problems at the frontier of voice AI.
Real impact: work on models used by real customers, not toy problems.
Significant compute to run ambitious experiments.
A small, fast-moving team where you’ll see how a frontier AI company is built from the inside.
Competitive internship allowance and benefits.
Interested?
We'd love to hear about you!
