Research Scientist, Pretraining, Gemma

Develop and evaluate next-generation frontier LLMs with an emphasis on pretraining and scalable multimodal data (speech, vision, text).
Design, implement, and assess training pipelines, architectures, and evaluation protocols to advance open-weights models.
Collaborate with researchers, engineers, and product teams to translate research into scalable, user-facing capabilities.
Tackle real-world applications across domains while ensuring safety, ethics, and robust evaluation practices.
Requirements: PhD (or equivalent) in AI; strong publication record; solid background in deep learning, NLP, CV, or speech; experience with JAX, TensorFlow, or PyTorch.
Nice-to-have: familiarity with VLM/VLA training and deployment; excellent communication and collaboration skills; passion for impactful, production-ready research.