Research Engineer, Data - Pika

Pika is looking for a staff-level Research Engineer, Data to architect and scale data engineering systems supporting model training for multimodal foundation models. This role involves owning large-scale data pipelines, curating diverse datasets, and optimizing data processing for training. Ideal candidates have 5+ years experience in data pipelines for ML, expertise in distributed systems like Spark and Ray, and strong programming skills.

• Take ownership of large-scale data pipeline architecture and implementation to support model training and research workflows for text, image, audio, and video datasets.
• Partner with research and engineering teams to curate, clean, and manage diverse, sensory-rich datasets for pre-training and mid-training of multimodal models.
• Develop strategies and tools for scalable data ingestion, labeling, filtering, augmentation, and storage.
• Optimize data processing, transformation, and delivery for large-scale distributed training pipelines.

• 5+ years of experience building and scaling data pipelines for machine learning applications at staff or lead engineer level.
• Strong background in data engineering and ML data curation for LLMs, VLMs, or other large-scale multimodal models.
• Expertise in distributed data systems (e.g., Spark, Hadoop, Ray) and efficient large dataset processing/ETL workflows.