Transformer Model Expert

Role overview Lead research and engineering work on transformer-based language models, improving their capabilities, efficiency, and readiness for use in production systems. The role spans architecture design, large-scale training and optimization, and responsible model alignment. It also includes collaborating across teams and sharing research through publications.

Responsibilities - Design, implement, and optimize transformer neural network architectures. - Lead research projects that advance language model capabilities. - Develop methods to improve model performance while reducing computational demands. - Apply alignment, safety, and ethical practices to model development. - Work with partner teams to bring research improvements into production. - Track relevant research and publish findings in leading conferences or journals.

Requirements - Master’s or Ph.D. in computer science, machine learning, or a related field. - At least three years of experience with transformer architectures such as GPT, BERT, or T5. - Strong NLP and deep learning expertise, demonstrated through publications or equivalent work. - Proficiency in Python and a deep learning framework such as PyTorch or TensorFlow. - Experience with distributed training and large-scale model training, fine-tuning, and optimization. - Understanding of attention, positional encodings, and sound machine learning engineering practices.

Nice to have - Experience with quantization, distillation, pruning, or other model compression methods. - Familiarity with reinforcement learning from human feedback, multimodal models, or efficient LLM inference and deployment. - Contributions to open-source machine learning projects.

Benefits and work setup - Full-time role based in Berkeley, California, or remote.