Data-Centric ML Research Engineer — Pre-Training

Liquid AI; United States; On-site; USD 150,000 - 210,000; Full time

Benefits offered by this job

Equity

Unlimited PTO

Refill Days company-wide

Comprehensive health coverage

Paid leave

Job summary

Liquid AI is seeking a seasoned pre-training data engineer to own data pipelines end to end. You will source, filter, deduplicate, and decontaminate diverse datasets and design curricula for pre-training and midtraining.

You’ll run experiments to quantify how data changes impact model quality, build synthetic data systems with LLMs and distillation, and ensure traceable data lineage for reproducible training runs.

Qualifications

• Deep, hands-on pre-training data work you did yourself: web-scale sourcing and curation, synthetic data generation, mixing and curriculum design, deduplication and decontamination.
• Ability to reason from a data decision to its effect on training: data interactions with model size, architecture, and training dynamics.
• Experiments you designed to test data changes with clear results and rationale.
• Production data pipelines you built and owned, not only research prototypes.
• Strong ML fundamentals demonstrated in Python (PyTorch preferred).
• 3+ years with MS, 1+ year with PhD, or 5+ years with BS.

Responsibilities

• Source web-scale and public datasets, and build pipelines to filter, deduplicate, and decontaminate them.
• Design data mixtures and curricula for pre-training and midtraining, and decide what goes into each run.
• Build synthetic data generation systems using LLMs, structured prompting, and distillation.
• Run proxy experiments and ablations to measure data changes on model quality and scalability.
• Build evaluation sets you can trust and maintain data lineage for reproducibility.

Skills

Research mindset

End-to-end ownership

Experiment design

Production data pipelines

Data sourcing and curation

Strong reasoning about data effects on

Education

Master's degree in a related field

PhD in a related field

Bachelor's degree in a related field

Tools

Python

PyTorch

Job description

Liquid AI is seeking a seasoned pre-training data engineer to own data pipelines end to end. You will source, filter, deduplicate, and decontaminate diverse datasets and design curricula for pre-training and midtraining.

You’ll run experiments to quantify how data changes impact model quality, build synthetic data systems with LLMs and distillation, and ensure traceable data lineage for reproducible training runs.