Staff Research Scientist, Generative AI & LLMs

Fireworks AI; United States; Sur place; USD 140 000 - 190 000; Plein temps

Résumé du poste

Fireworks AI is seeking a researcher finishing a PhD to join the Research team as a Member of Technical Staff. The role targets advancing LLMs and multimodal systems, with a path from research to production in a fast-moving environment.

You will work with a senior researcher mentor, tackling real research problems from day one and collaborating with experts across deep learning, distributed systems, and optimization to ship models for leading customers in weeks rather than years.

Qualifications

• PhD completed within a reasonable timeframe or anticipated by Dec 2026 in a relevant field.
• Strong research background demonstrated by dissertations, publications, or projects.
• Hands-on experience with ML frameworks (PyTorch, JAX, TensorFlow).
• Proficient in Python; C/C++ or similar languages.

Responsabilités

• Finish PhD research with impactable production relevance, not just publication.
• Push boundaries of generative AI, improving LLMs and multimodal systems.
• Work with a mentor and researchers from deep learning, systems, and optimization.
• Conduct foundational research to advance LLMs and multimodal capabilities.
• Design and evaluate novel model architectures and training methods.
• Collaborate with engineering to transition research prototypes to production.
• Analyze results, identify bottlenecks, and iterate to improve model quality.
• Bring doctoral insights into internal research agenda.

Connaissances

Python programming

Analytical thinking

Communication skills

Research experience

Formation

PhD in CS/ML/related field

Outils

PyTorch

JAX

TensorFlow

Description du poste

Fireworks AI is seeking a researcher finishing a PhD to join the Research team as a Member of Technical Staff. The role targets advancing LLMs and multimodal systems, with a path from research to production in a fast-moving environment.

You will work with a senior researcher mentor, tackling real research problems from day one and collaborating with experts across deep learning, distributed systems, and optimization to ship models for leading customers in weeks rather than years.