Perception Data Engineer, ANYbotics

• 3+ years engineering experience building production data pipelines or ETL systems
• Strong Python scripting and engineering skills (pandas, pyarrow, boto3 or equivalent)
• Experience with dataset versioning or large-file management (DVC, Git-LFS, or similar) and cloud object storage (S3)
• Familiarity with annotation tooling and workflows for image data (CVAT / Label Studio)
• Basic understanding of ML training data needs (batching, sharding, augmentation integration)
• Prior work supporting computer-vision teams (image pipelines, preprocessing, TFRecord or custom dataset formats)

Desirable

• You’ll build and operate the data plumbing that our perception models need: ingestion, versioned storage, ETL, labeling integration, and reliable production pipelines for training and inference
• Design, build and maintain scalable data pipelines and ETL workflows that ingest raw images, sensor metadata, and labels (both real and synthetic)
• Implement dataset versioning, schema management, and reproducible data snapshots to support experiments and audits
• Integrate annotation tools (CVAT / Label Studio), manage labeling workflows and quality-control tooling, and support label QA processes
• Build data validation and monitoring checks (file integrity, label sanity, distribution drift alerts) and automate remediation where possible
• Provide clean, ready-to-use datasets and data loaders for ML engineers; optimize data access patterns for training (sharding, caching, prefetching)
• Collaborate with MLOps to automate scheduled retraining triggers and with Synthetic Data Engineer to merge synthetic data streams