Applied AI/ML Scientist: Build & Scale Custom LLMs

Cerebras; Abu Dhabi; Presencial; AED 661.000 - 1.028.000; Jornada completa

Descripción de la vacante

We are seeking an Applied AI Scientist in the FieldML team at Cerebras to develop and customize large language models and other large-scale deep learning systems for customer-specific problems. You will bridge research and production, harnessing the Cerebras Wafer-Scale Engine to accelerate AI initiatives and deliver measurable customer value.

The role involves end-to-end training, model customization, and collaboration with customers to scope engagements, design training recipes, and optimize

Formación

• Master’s or PhD in Computer Science, Machine Learning, or related field.
• Proven track record training/fine-tuning large models (1B+ parameters).
• Strong Python and PyTorch proficiency; experience with distributed training.
• Excellent communication and ability to translate complex results for diverse audiences.

Responsabilidades

• Collaborate with customers to scope AI engagements and define success metrics.
• Architect and implement end-to-end training recipes for custom models.
• Design adaptation strategies including SFT, RLHF, or DPO as needed.
• Own training pipelines, data preprocessing, tokenization, and hyperparameter tuning.
• Scale workloads across Cerebras clusters and contribute to tool-use capabilities.

Conocimientos

Deep learning expertise

Model training (1B+ parameters)

Python

PyTorch

Distributed training

Communication skills

Educación

Master’s or PhD in CS/ML

Herramientas

PyTorch

Distributed training frameworks

Descripción del empleo

We are seeking an Applied AI Scientist in the FieldML team at Cerebras to develop and customize large language models and other large-scale deep learning systems for customer-specific problems. You will bridge research and production, harnessing the Cerebras Wafer-Scale Engine to accelerate AI initiatives and deliver measurable customer value.

The role involves end-to-end training, model customization, and collaboration with customers to scope engagements, design training recipes, and optimize