This description is a summary of our understanding of the job description. Click on 'Apply' button to find out more.

Role Description

As a Geospatial Data Platform / Label Ops Engineer on the AI/Advanced Engineering team, you’ll own the imagery and labeling data plane behind SkyFi’s near-real-time satellite analytics, making diverse partner imagery fast to ingest, consistent to use, and reproducible end-to-end. You’ll build and operate scalable pipelines to normalize and catalog imagery across many sensors/providers, deliver high-performance tiling/chipping and retrieval services for training and inference, and implement dataset + label versioning and lineage so every model output and evaluation result can be traced back to the exact data used. You’ll define and maintain our labeling pipeline with QA/adjudication and auditability. Working closely with CV and runtime owners, you’ll ship self-serve data products that speed up iteration and improve accuracy. This is a high ownership position where you’ll be a cornerstone member of a team that is empowering the future of Geospatial AI.

Qualifications

• Demonstrated experience building geospatial imagery pipelines at scale (raster workflows, tiling/chipping, handling heterogeneous sensors/metadata).
• Strong data engineering fundamentals: idempotency, backfills, observability, SLAs, schema evolution, and production reliability.
• Experience building internal data APIs/SDKs and treating data as a product.
• Hands-on experience with labeling workflows or data QA at scale (vendor coordination, task design, QA/adjudication mechanics).
• Ability to collaborate tightly with CV/eval owners to translate failure modes into actionable data/labeling pipelines.

Requirements

• Own the imagery data plane: ingest, normalize, catalog, and serve imagery + metadata across diverse sources for near-real-time and batch workloads.
• Build and operate tiling/chipping + retrieval services optimized for training and NRT inference (spatial/temporal indexing, caching, precompute, and latency SLAs).
• Implement dataset and label versioning + lineage so every model run / evaluation can be reproduced.
• Build and run label ops workflows: task generation, QA, adjudication, gold-check insertion, audit-ability, throughput tracking.
• Create data products for internal consumers (APIs/services) that let CV engineers self-serve imagery chips, labels, and eval sets.
• Build robust backfill/reprocessing pipelines (idempotent, observable, safe incremental recompute) to support new analytics and changing requirements.
• Establish data health monitoring (freshness, completeness, corruption, sensor distribution drift, metadata validation) with alerts and dashboards.
• Partner with evaluation and runtime owners to close the loop of failure buckets -> labeling requests -> dataset versions -> retraining/eval.
• Partner with computer vision researchers to define image and label strategies for new projects.
• Responsible for making sure everyone has the images/data/labels they need.

Benefits

• Be well compensated. Possibility for equity.
• Receive best-in-class benefits, including premium medical, dental, and vision coverage and 20 days paid time off.
• Play a critical role in building a market-changing product in the exciting realm of Space.
• Thrive in a fast-paced, dynamic environment that rewards initiative, innovation, and getting things done.

Salary Band

$180,000–$220,000 USD base salary