Principal AI Engineer, Machine Learning Operations

BlackSky is a real-time intelligence company providing satellite imagery, automated analytics, and high-frequency monitoring through a space-based intelligence platform. The Principal AI Engineer, Machine Learning Operations will serve as the technical authority for designing, scaling, deploying, and automating computer vision and machine learning architectures for real-time geospatial analytics, while leading model development, production integration, and technical mentorship.

Responsibilities:

• End-to-End Model R&D: Leading full-lifecycle model engineering from initial problem formulation and dataset curation to training, validation, deployment, and error analysis
• Model Robustness & Domain Adaptation: Driving model resilience against challenging real-world satellite conditions, including off-nadir angles, low light, seasonal variations, weather distortions, dense environments, and cross-sensor domain shifts
• Production Integration: Ensuring models are reliably trained, versioned, deployed, monitored, and maintained in production
• Technical Leadership & Strategy: Communicating technical progress, risks, and stack to cross-functional leadership and Product partners while contributing to proposals and white papers
• Mentorship & Engineering Culture: Providing technical mentorship, spearheading design reviews and experiment planning, and fostering a culture of technical excellence across the engineering team
• Other job-related duties as assigned

Requirements:

• Minimum of 12 years of hands-on software engineering experience, including at least four years developing computer vision or machine learning solutions for geospatial, remote sensing, or similarly complex imagery
• Bachelor's degree in computer science, engineering, mathematics, or a related quantitative field, or equivalent practical experience
• Expert-level proficiency in Python and demonstrated experience developing deep learning systems using PyTorch or comparable modern ML frameworks
• Proven experience designing, training, evaluating, and deploying production-grade computer vision models for tasks such as object detection, segmentation, change detection, image classification, or time-series analysis
• Strong experience working with remote sensing imagery and the unique challenges associated with geospatial data, including varying spatial resolutions, viewing geometries, sensors, and environmental conditions
• Hands-on experience building geospatial data and CV pipelines using tools such as Rasterio, GDAL, GeoPandas, Shapely, xarray, Zarr, or similar technologies
• Demonstrated experience taking CV capabilities from experimentation through production, including model evaluation, deployment, and performance optimization
• Experience developing and operating ML workloads in AWS, including GPU-based training and inference
• Strong technical communication skills with the ability to communicate architecture decisions, experimental results, tradeoffs, risks, and recommendations to both technical and non-technical stakeholders
• Demonstrated ability to provide technical leadership and mentorship across engineering teams without requiring direct management authority
• The work this role supports requires holding US citizenship
• Master's degree or Ph.D. in computer science, engineering, mathematics, machine learning, remote sensing, or a related quantitative field
• Twelve or more years of hands-on software engineering, computer vision, or machine learning experience
• Experience working with imagery from multiple commercial and government remote sensing sources, such as BlackSky, Sentinel, Airbus, Planet, or WorldView
• Demonstrated experience optimizing ML systems for production constraints, including inference latency, throughput, memory utilization, GPU efficiency, and cost
• Experience rapidly developing prototypes or proof-of-concept applications and communicating results through customer demonstrations, proposals, white papers, or other technical materials
• Strong experience with modern MLOps practices, including experiment tracking, dataset and model versioning, reproducibility, automated evaluation, and production monitoring
• Experience applying foundation models, vision-language models (VLMs), multimodal approaches, or self-supervised learning to geospatial or remote sensing problems
• Experience adapting models across sensors, geographic regions, resolutions, collection conditions, or other significant domain shifts
• you must be a U.S. citizen, lawful permanent resident of the U.S., protected individual as defined by 8 U.S.C. 1324b(a)(3), or eligible to obtain the required authorizations from the U.S. Department of State