Agentic Data Understanding
Core
Design and build AI-driven systems to automatically generate high-quality annotations at scale for autonomous construction robots, transforming raw sensor data into structured, semantically rich data.
Role type
Senior IC Agentic Data Understanding Engineer
Builds
Auto-labelers, orchestration pipelines, quality harnesses, and agentic workflows for robot autonomy
Domain
Robotics, Autonomous Vehicles, Construction Technology
Deliverable
production ML models
Required skills
ML engineering, data engineering, applied AI, Python, complex pipeline development, ML model inference at scale, annotation modalities (2D bounding boxes, 3D cuboids, semantic segmentation, event/context labels), quality/cost tradeoff analysis, metric instrumentation, cross-functional collaboration
Preferred skills
Model fine-tuning, agentic or LLM-orchestrated pipelines, VLMs for zero-shot/few-shot annotation, robotics data formats (LiDAR, cameras, IMU), annotation platforms, ontology/taxonomy design
Technologies
Python, VLMs, LiDAR, cameras, IMU
Responsibilities
Design hybrid cascading auto-labeling pipelines; develop annotation harnesses for quality/cost assessment; productionize annotation workflows; build and maintain the Annotation Orchestrator; implement scheduling logic for parallel workflows; build agentic annotation workflows for self-diagnosis and human-in-the-loop review; develop tooling for annotation versioning; define and instrument metrics for quality/cost optimization
Seniority
Senior, hands-on IC