Data Scientist
Description
MSG is looking for a Data Scientist to join a groundbreaking Capture and Imaging Systems department based in Burbank, CA. As the Capture and Imaging Systems team for the MSG Sphere projects, we strive to create tools for world-class content at an unprecedented scale on the largest LED screens ever made.
The Data Scientist works in close coordination with the Capture and Imaging Systems, post-production and capture departments to develop solutions and tools to automate the content creation workflow and enhance image quality to the highest standards for MSG Sphere. Candidates must have experience working on machine learning for imaging applications. Challenges will include multicamera fusion, content segmentation, de-mosaicking, super resolution, temporal stability, temporal up-sampling, denoise, artifact correction, content aware corrections, color enhancement, etc.
The Capture and Imaging Systems team is tasked with creating and maintaining hardware and software tools for preview, capture, ingest, transcode, color science, review, QC, and final delivery. Within this scope, the Data Scientist will be focused on improving final image quality and automating processes leading up to final output. Additionally, the Data Scientist will also adapt some solutions for real time use cases.
The resolution, field of view, and sheer scale of the Sphere venues means that the image quality for Sphere content creation is critical. The Data Scientist must be able to work closely with internal users, iterate quickly, prioritize, and develop models that can be deployed to meet the production goals.
SPECIFIC FUNCTIONS/ACTIVITIES SUMMARY
Create and document image collection procedures for machine learning
Develop machine learning based solutions for de-mosaicking and super resolution
Develop content segmentation tools for layering machine learning models
Architect and develop adversarial or similar network approaches to image enhancement and stitching
Develop content aware fill and continuity correction algorithms
Develop temporally stable inference tools
Create transfer learning workflows for adaptable image enhancement
Optimize techniques for cinema quality results
Write and review invention disclosures for developed algorithms and work with patent attorneys to secure intellectual property
Analyze the performance of the image enhancement and automation tools
Bachelor’s degree or equivalent in data science, Master’s degree preferred
7 years of experience in data science
In-depth knowledge of modern deep learning techniques for image processing
Familiar with one of the deep learning frameworks, Tensorflow, Pytorch or MXNet
Familiarity with heterogeneous computing systems (OpenCL, CUDA, TPUs)
Experience in Generative Adversarial Networks or similar feedback strategies
Experience in transfer learning and fine tuning.
Experience in computer vision techniques including image/video processing, object detection, classification and tracking.
Experience with one or more general purpose programming languages including but not limited to: C/C++ and Python
Broad grounding in image quality fundamentals and theory, including optics, image sensors, camera systems, lighting, and display technologies
Experience deploying models to Nvidia RTX, Jetson or similar architectures.
This Job is no longer active!