Andrew Noble Email and Phone Number

• Led development, maintenance, and strategic planning for a Software Development Kit of re-useable, fast, scalable components for Knowledge Base Management, Data Engineering, and AI Modeling. Deployed in production environment and running bug-free for >1yr. Enabled 5x reduction in delivery time on client projects. Coordinated enhancements from 10+ contributors and built cross-functional community of 25+ users.• Led early adoption of Generative AI methodology for multimodal EHR data. Implemented Transformer models from scratch. Built well-documented and tested software package enabling cost-effective multi-GPU training and inference on AWS Batch with logging to MLFlow. • Demonstrated Generative AI capability to reliably infer unknown, time-dependent Clinical features (e.g., Performance Status) from commonly available Administrative Claims data, with the potential to dramatically scale and accelerate pre-screening of common Inclusion/Exclusion Criteria for Clinical Trials. Published ASCO 2023 Abstract, Dr. Ravi Parikh co-author.

• Led AI NLP development of online extraction of symptoms from clinical notes and surveys to prioritize patients for follow-up. Named-Entity-Recognition Transformer model prototyped on a cohort of 165k oncology patients.• Led AI Imaging platform development, semi-automating organ and tumor segmentation workflows with state-of-the-art Computer Vision algorithms. Rapidly scaled interest among radiology labs. • Developed Data Engineering pipeline for conversion of site-specific EHR to the OMOP Common Data Model. Handoff scaled to >2M patient records, enabling multi-site collaboration on rare cancer screen.

• Led AI Longitudinal Causal pipeline development of oncology-specific disease progression and risk models.• Led Data Engineering efforts, performing ETL on multimodal EHR for client projects and product development.• Informed vision for platform development pipelines and algorithms. Demonstrated business value of state-of-the-art Deep Learning frameworks and Variational Inference methods.

• Consulted with Data Science team at Merck to prototype a web app for automating information sharing among patients and caregivers in online health communities. • Cleaned 6,500 Reddit posts on Dementia and Alzheimer's forum using Jupyter, Pandas, RE, and NLTK Python packages for NLP. Stored posts in PostgreSQL database. • Discovered a 3-topic Latent Dirichlet Allocation model, trained with Gensim, describing >90% of the unstructured data. • Designed a web application with Bootcamp, D3.js, and Flask, served up by Gunicorn on an AWS EC2 instance, that classifies user input and recommends related content.

• Forged research team and led grant writing effort to win a $600,000 interdisciplinary award from National Science Foundation with joint support from Directorates for Mathematical and Physical Sciences and Biological Sciences. An additional 4-year, $1M non-traditional NSF Award supports on-going research• Discovered a robust theoretical correspondence between biological oscillator networks and magnets with observational evidence found in a pistachio orchard.• Developed custom C/C++ software to distribute Large-scale Monte Carlo Simulations over 192 cores of a Linux cluster. Bootstrapped statistical errors in over 100GB of synthetic data.• Created animations of emergent fractal patterns in oscillator networks using Matplotlib.• Co-organized two multidisciplinary working groups on Physics and Biology at Santa Fe Institute.

• Discovered new mathematical method for analyzing Multivariate Stochastic Processes enabling fast characterization of steady states describing evolutionary fixed points.• Designed and deployed a plug-in menu of graph theory algorithms for QGIS, an Open Source Geographic Information System, using Qt, C++ Boost Graph Library, and PostgreSQL.

• Discovered potential for Higgs Boson self-coupling interactions at next-generation particle colliders to probe origins of matter/anti-matter asymmetry in universe. Searching for this signature now informs not only theoretical research into new particles and forces but also the engineering and design of future experiments.• Built Mathematica notebooks to automate calculation of Feynman Diagram estimates for particle interactions and decay rates, simplifying thousands of pages of algebra into a single line and generating actionable predictions. • Constrained theoretical results for physics beyond the Standard Model with data from both terrestrial particle colliders, such as the Large Electron Positron experiment at CERN, and astrophysical observatories, such as Planck.

• Named teacher of the month in January 2001 for teaching students HTML and web development. • Founded a mentorship program in collaboration with 12 students from Johns Hopkins University as well as the Chair of the Hopkins Physics Department, Richard Henry.• Led a team of students to a 2nd place finish in a robotics competition sponsored by IEEE. In a field of 26 teams, we were the only non-magnet Baltimore City Public Schools team to compete.