Selected work
Production machine learning in healthcare and beyond: the platforms, pipelines, and products Randolph Schilke has built, how they work, and what they changed. For the full timeline, see the résumé.
Blue Cross Blue Shield of North Carolina, via CuraCor Solutions
Architecting an enterprise hub-and-spoke MLOps platform on AWS CDK: a shared hub, a versioned construct library, and a template that generates ML projects.
Blue Cross Blue Shield of North Carolina, via CuraCor Solutions
Designed and own the AWS CDK platform that trains, promotes, serves, and monitors 37 clinical prediction models across 8 lines of business at Blue Cross NC.
U.S. Department of Veterans Affairs, Office of Connected Care, via InnoVet Health
Built the VA's first production LLM pipeline: clinical note processing on Azure Databricks, still in automation with an estimated $5M+ in annual savings.
U.S. Department of Veterans Affairs, Office of Connected Care, via InnoVet Health
Maintained and extended the VA's most mature predictive model, which flags discharged veterans at highest risk of 30-day readmission, with MLflow-based MLOps.
Mercy, one of the largest U.S. health systems
Led a 10-person data science team and stood up Mercy's first MLOps pipeline on Azure Databricks, including its first real-time models integrated with EPIC.
Schilke App Solutions LLC
Built a robot platform and iOS/Android app: WebRTC remote driving, an on-device LLM on a Jetson Orin Nano, and YOLO object detection on a Raspberry Pi 5.
Schilke App Solutions LLC
Shipped an iOS app that identifies fossil shark teeth in real time with 18 on-device Core ML classifiers trained on 100,000+ labeled images.
Florida Department of Health
Applied text mining to EMS clinical notes to sharpen Florida's opioid overdose surveillance, and owned the state's CDC data deliverables for the program.
Contact
I'm open to Senior and Staff AI engineering, ML platform, and data science leadership roles, and to healthcare AI consulting. Email is the fastest way to reach me, and I reply personally.