Foundation Model Data Engineer: Petabyte-Scale Data Pipelines

Sciforium; San Francisco (CA); On-site; USD 180,000 - 240,000; Full time

Benefits offered by this job

Medical insurance

401k plan

Lunch, snacks and beverages

Flexible time off

Competitive salary and equity

Job summary

Sciforium is seeking a highly technical Foundation Model Data Engineer to own end-to-end data lifecycle for large-scale AI models. You will architect dataset strategies, curate petabyte-scale data, and design post-training alignment datasets to improve model reasoning and safety.

Ideal candidates have 5+ years in data science or ML, deep Python skills, and experience with petabyte-scale datasets, roadmapping data pipelines, and RLHF/DPO data workflows.

Qualifications

• 5+ years of industry experience in Data Science or ML focused on datasets for foundation models.
• Expert-level Python with multiprocessing, multithreading, and memory management for large-scale data tasks.
• Petabyte-scale datasets used for training production-grade LLMs or Large Vision Models.
• Dataset reconstruction from raw web crawls (e.g., Common Crawl) and domain data.
• Hands-on RLHF, DPO, and multi-turn instruction-following data creation and labeling workflows.
• Mastery of data-at-scale frameworks like Spark, Ray, or high-performance data-loading formats.

Responsibilities

• Own end-to-end data lifecycle from raw web-scale crawling to alignment datasets.
• Design taxonomies, filtering, and post-training pipelines for model behavior.
• Develop data processing scripts with high throughput for ingestion and transformation.
• Lead development of post-training datasets including SFT instructions and multi-turn data.
• Drive multimodal data acquisition and processing for vision and video modalities.
• Profile data to identify biases and gaps to balance model training data.

Skills

5+ years experience in datasets

Python expert

multimodal data pipelines

data profiling & analysis

high-performance data processing

statistical data analysis

Tools

Spark

Ray

WebDataset

Parquet

Job description

Sciforium is seeking a highly technical Foundation Model Data Engineer to own end-to-end data lifecycle for large-scale AI models. You will architect dataset strategies, curate petabyte-scale data, and design post-training alignment datasets to improve model reasoning and safety.

Ideal candidates have 5+ years in data science or ML, deep Python skills, and experience with petabyte-scale datasets, roadmapping data pipelines, and RLHF/DPO data workflows.