Skip to main content
TALENT247 // SPECIALIST AI EXECUTIVE SEARCH & TECHNICAL RECRUITMENT
OPS · MLOps & Platform

Senior AI Data Platform Engineer, Multimodal Curation

Seattle, WA · Full-time · Hybrid

AI Specialism

MLOps & Platform

Workplace

Hybrid

Location

Seattle, WA

Compensation

$190k – $270k

annual package

Role Dossier & Specification

REF: SENIOR-AI-DATA-PLA

THE CLIENT & MANDATE

Our client is a flagship AI data infrastructure and post-training company powering the multimodal, video, and physical-AI datasets behind leading frontier model labs and robotics pioneers. From their Seattle engineering center, their Multimodal Data Platform group processes hundreds of petabytes of raw video, 3D LiDAR point clouds, multi-view audio, and bimanual robot teleoperation trajectories.

They have engaged Talent247 to appoint a Senior AI Data Platform Engineer for Multimodal Curation in Seattle, WA. The successful candidate will build distributed GPU-accelerated curation, semantic deduplication, and synthetic auto-labeling engines that transform noisy raw sensor captures into frontier-grade pre-training and post-training datasets.

KEY RESPONSIBILITIES

  • Architect horizontal Ray Data and Spark pipelines running distributed GPU inference (VLM captioning, optical flow filtering, 3D bounding-box tracking, audio diarization) across petabyte multimodal lakes.
  • Build high-throughput columnar storage and random-access indexing layers on Lance, Apache Iceberg, and Parquet enabling sub-second spatio-temporal slice queries across billions of multimodal frames.
  • Implement semantic deduplication, aesthetic/motion quality scoring, and MinHash/embedding clustering algorithms that maximize downstream foundation model sample efficiency.
  • Engineer synchronized multi-sensor ingestion pipelines validating sub-millisecond timestamp alignment and camera-LiDAR-proprioception calibration across robotics and autonomous fleet uploads.
  • Design human-in-the-loop quality-routing engines that direct high-entropy model disagreements to expert annotators while auto-accepting high-confidence teacher labels.
  • Optimize spot-GPU scheduling, NVDEC hardware video decoding, and object-store multipart streaming to cut per-million-clip processing costs by 50%+.

TECHNICAL ENVIRONMENT & OFFERING

Stack: Python, Rust/C++, Ray Data, Apache Spark, Lance, Iceberg/Parquet/Arrow, PyTorch, vLLM, Kubernetes, S3/GCS. Workplace: Hybrid in Seattle, WA. Compensation: $190,000 – $270,000 base salary plus equity ownership and full US health & retirement benefits.

Core Requirements

ESSENTIAL QUALIFICATIONS

  • 5+ years of distributed data platform or ML infrastructure engineering experience handling multi-terabyte to multi-petabyte unstructured or multimodal datasets.
  • Production mastery of Ray Data, Apache Spark, or Dask in Python, paired with strong systems fluency in Rust, Go, or C++.
  • Hands-on experience with multimodal columnar table formats (Lance, Iceberg, Delta, Parquet) and GPU-accelerated batch inference orchestration.
  • Solid grasp of vector indexing, approximate nearest-neighbor clustering, and object-storage throughput optimization.
  • Track record owning end-to-end reliability, cost telemetry, and schema evolution for mission-critical data pipelines.

PREFERRED / BONUS EXPERIENCE

  • Direct experience processing robotics trajectory formats (RLDS, LeRobot, MCAP/ROS2 bags), 3D point clouds, or long-form multi-stream video.
  • Familiarity with hardware-accelerated video/image decoding (NVDEC, DALI) and VLM-based synthetic recaptioning pipelines.
  • Open-source contributions to Ray, Lance, Arrow, or multimodal dataset toolkits.

Position Telemetry

Location Seattle, WA (Hybrid)
Department MLOps & Platform
Employment type Full-time
Workplace type Hybrid
Salary range
$190k – $270k annual package