混元多模态数据开发工程师（北京）
Core
Design and develop multi-modal AI data pipelines for voice, image, video, and 3D data to support large-scale training dataset production.
Role type
Senior multi-modal data engineering IC
Builds
High-throughput data storage, indexing systems, and data processing pipelines for AI training
Domain
Artificial Intelligence / Multi-modal Data Engineering
Deliverable
production ML models
Required skills
Python, Go, C++, Linux, Kafka, Pulsar, RocketMQ, Airflow, Celery, OSS, S3, HDFS, Ray, Spark, Flink
Preferred skills
Voice preprocessing (noise reduction, VAD, ASR), Image/Video processing (OpenCV, Pillow), 3D data formats (Point Cloud, Mesh, NeRF), Automated data quality evaluation, Data synthesis, AIGC-assisted data production
Responsibilities
Design multi-modal AI data pipelines; Build efficient data preprocessing pipelines; Collaborate with algorithms/researchers on data construction and alignment; Optimize pipeline stability and performance; Explore automated data quality and synthesis technologies