Senior Multimodal ML Data Engineer

LVT; Seattle, Northern (WA, KY); Hybrid; USD 172,000 - 221,000; Full time

Benefits offered by this job

Comprehensive health

Dental and vision coverage

401k match

Flexible PTO

Job summary

LVT is seeking a Staff Data Engineer to own the data flywheel end to end, from raw edge telemetry and video to labeled training datasets and evaluation sets. You will partner with AI/ML teams, own dataset schema, versioning, and governance, and lead data-engineering standards without formal management requirements.

You will design scalable lakehouse architectures, integrate PyTorch/Lightning data loaders, and enable self-serve access to data for research while ensuring privacy and provenance

Qualifications

• 8+ years building and operating large-scale data pipelines and data-lake or lakehouse systems in production
• Experience with model training data, labels, and evaluation datasets
• Knowledge of medallion-style architectures and modern table formats
• Experience with multimodal data (video, image, sensor) and scalable storage/access patterns
• Strong Python and Spark experience; familiarity with PyTorch/Lightning dataloaders

Responsibilities

• Own end-to-end Data Flywheel from raw telemetry/video to labeled training data
• Build and maintain pipelines to standardize data into a single schema
• Manage labels/annotations lifecycle and dataset versioning
• Own evaluation/benchmark datasets and ensure stable, comparable model comparisons
• Enforce governance, lineage, and privacy controls across datasets and pipelines
• Mentor ICs and raise data-engineering standards across the team

Skills

Data engineering

ML data

Lakehouse architecture

Multimodal data

Python

PyTorch/Lightning

Education

Bachelor's/Master in CS/Engineering

Tools

Spark

Iceberg

Parquet

Delta Lake

Kafka

Job description

LVT is seeking a Staff Data Engineer to own the data flywheel end to end, from raw edge telemetry and video to labeled training datasets and evaluation sets. You will partner with AI/ML teams, own dataset schema, versioning, and governance, and lead data-engineering standards without formal management requirements.

You will design scalable lakehouse architectures, integrate PyTorch/Lightning data loaders, and enable self-serve access to data for research while ensuring privacy and provenance