PhD Research Intern (Spring/Summer 2025)

As a PhD Research Intern at Cartesia, your main responsibilities include implementing new model architectures, rapidly iterating on experiments and ablations, and building infrastructure for large multimodal datasets. You'll collaborate closely with various stakeholders to enhance the quality and capabilities of our pretrained models and stay informed about the latest research developments in the field.

Candidates applying for the PhD Research Intern position at Cartesia should have a deep understanding of machine learning principles, particularly in sequence modeling and generative models. Proficiency in Python and frameworks like PyTorch is essential, as is experience with training large-scale models. Additionally, familiarity with efficiency trade-offs in model architecture design for GPU accelerators, as well as having main-track research publications at top-tier conferences, will significantly strengthen your application.

A successful PhD Research Intern at Cartesia should be proficient in Python, with a strong working knowledge of PyTorch or similar frameworks. Familiarity with tensor programming and prior experience in optimizing model inference using CUDA or Triton is also advantageous. The role involves navigating complex codebases, making strong programming skills crucial for success.

Cartesia fosters a vibrant and collaborative work culture where all team members are encouraged to learn from one another. The office environment in San Francisco is designed for in-person teamwork and innovation, allowing interns to engage closely with experienced researchers. The company values execution speed and helps support everyone in achieving their goals with an open and inclusive approach.

Yes! PhD Research Interns at Cartesia receive several benefits, including lunch, dinner, and snacks provided at the office. There is also relocation assistance for interns moving to San Francisco, alongside a welcoming atmosphere that fosters teamwork and quick learning. With perks like these, interns can fully immerse themselves in their roles without worrying about daily essentials.

When answering this question, focus on specific architectures you have worked with, detailing any unique challenges and how you addressed them. Emphasize your understanding of the impact of architecture design on model performance and your ability to stay updated with current research.

Your response should illustrate your systematic approach to experimentation, including how you design experiments, manage resources, and analyze results. Mention any tools or methodologies you use to keep your experiments organized, as well as how you communicate findings with your team.

Talk about specific projects or research where you worked with multimodal models, underscoring the challenges you faced and the solutions you devised. Highlight your understanding of integrating various data types and how they enhance model performance.

Highlight your grasp of sequence modeling, including relevant projects where you've applied techniques such as RNNs, LSTMs, or Transformers. Discuss how you analyze sequential data and how such models impact predictive accuracy and efficiency.

Elaborate on your strategies for keeping up with AI research, such as following key conferences, reading academic journals, and participating in online forums. Emphasize any specific communities or researchers that influence your work and thinking.

When answering this, address fundamental efficiency trade-offs in model architecture design, such as memory bandwidth versus computational speed. Discuss specific examples in your past work where these considerations led to notable improvements in performance.

Share a specific project where you pre-trained a large-scale model, detailing your role, the methodologies you employed, and the eventual outcomes. Highlight any insights gained from this experience that would be relevant to the PhD Research Intern position.

Discuss your knowledge of state space models and their applications. If you have prior experience in this area, share specific instances where you advanced this concept or implemented such models in practice, showcasing their relevance to Cartesia's work.

Explain your methods for prioritization, including how you assess project importance and deadlines. Discuss any tools or frameworks you use to keep track of tasks and how you maintain communication with your team to manage expectations.

Talk about specific collaboration tools that you are comfortable using, such as version control systems like Git or project management software. Highlight your experience in fostering an open and inclusive team environment, illustrating how you contribute to effective communication and collaboration.