Applied ML research scientist at Cerebras who builds and scales ML workflows for LLM pretraining, fine-tuning, and reinforcement-learning post-training — day to day this means building evaluation and data pipelines, debugging across the ML stack, and optimizing training and inference. Core stack: Python, PyTorch, and transformer-based deep learning.
You will implement scalable ML workflows for LLM pretraining, fine-tuning, and reinforcement learning post-training. You will build evaluation and data pipelines, debug issues across the ML stack, optimize training and inference, and contribute maintainable infrastructure code.
Responsibilities
Apply post-training techniques to improve model performance
Build and maintain model evaluation pipelines
Debug data pipelines, training jobs, model outputs, and lower-precision computation
Translate ML ideas into scalable implementations
Design and scale ML pipelines for pretraining, fine-tuning, and alignment
Generate, filter, and use large datasets and synthetic data
Optimize training and inference workflows for performance, efficiency, and reliability
Contribute maintainable code to shared ML infrastructure