Research Engineer (Foundation Model) New Grad

We are seeking a highly skilled Research Engineer with extensive experience in training Generative AI models. As part of our research team, you will play a key role in building state-of-the-art multimodal foundation models and managing large-scale training runs on thousands of GPUs. Your expertise will directly impact the performance, scalability, and efficiency of our next-generation AI technologies.

Key Responsibilities

• Lead and contribute to groundbreaking research in multimodal foundation models.
• Design, develop, and experiment with innovative algorithms, architectures, and techniques to enhance model performance and scalability.
• Optimize models for production environments, focusing on computational efficiency, throughput, and latency while maintaining accuracy and robustness.
• Analyze and manage large-scale data clusters, identifying inefficiencies and bottlenecks in training pipelines and data loading processes.
• Collaborate with cross-functional teams, including data, applied research, and infrastructure teams, to drive impactful projects.

Qualifications

• Technical Expertise:
  • Demonstrated strong engineering skills in Python and PyTorch.
  • Hands-on experience building machine learning models from scratch using PyTorch.
  • Familiarity with generative multimodal models such as Diffusion Models and GANs.
  • Solid understanding of foundational deep learning concepts, including Transformers.
• Preferred Experience:
  • 1 year+ industrial or academic lab experience.
  • Experience working with large distributed systems involving 100+ GPUs.
  • Proficiency with Linux clusters, systems, and scripting.

Note: This role is open to recent graduates.

Compensation

The salary range for this position in California is $160,000–$200,000 per year. The final offer will be based on job-related expertise, skills, candidate location, and experience. Additionally, we provide competitive equity packages in the form of stock options and a comprehensive benefits plan.