Staff - Pretraining Data & Pipelines (Equity)

Digitailor; United States; In loco; USD 300.000 - 400.000; Tempo pieno

Vantaggi offerti da questo lavoro

Equity grant

Competitive benefits

Descrizione del lavoro

Mirendil Mirendil is seeking a data-driven engineer to define training data inputs and manage large-scale pretraining experiments. You will orchestrate data pipelines, ablations, and parallelized runs, collaborating with researchers on model architecture and training runs.

You will justify data choices with rigorous experiments, ensuring data provenance and reproducibility as the corpus scales. A competitive equity grant accompanies a base salary in a high-performance, research-focused

Competenze

• Shipped a data change into a large pretraining run.
• Built data pipelines at the scale of trillions of tokens with Spark, Ray, Beam.
• Strong Python; experience with PyTorch or JAX for training experiments.
• Able to justify ablations and experimental results with writing.

Mansioni

• Set the direction for what data the models train on.
• Design ablations and data scaling experiments and justify choices.
• Direct agents to run experiments and pipelines in parallel.
• Maintain provenance and reproducible data versions for training runs.

Conoscenze

Python

PyTorch

JAX

Spark

Data pipelines

Experiment design

Data deduplication

Data scaling

Strumenti

Spark

Ray

Beam

Descrizione del lavoro

Mirendil Mirendil is seeking a data-driven engineer to define training data inputs and manage large-scale pretraining experiments. You will orchestrate data pipelines, ablations, and parallelized runs, collaborating with researchers on model architecture and training runs.

You will justify data choices with rigorous experiments, ensuring data provenance and reproducibility as the corpus scales. A competitive equity grant accompanies a base salary in a high-performance, research-focused