[{"content":"","externalUrl":null,"permalink":"/categories/","section":"Categories","summary":"","title":"Categories","type":"categories"},{"content":"I have worked on health-system data at Michigan Medicine since 2013. That work started as Splunk architecture and data science, and since 2026 it has been security data engineering in Information Assurance. Before that I did machine-learning research on budgeted and active learning. I also teach in the University of Michigan\u0026rsquo;s master\u0026rsquo;s program in data science.\nWhat I am working on now # Security data and detection. Splunk and Cribl pipelines, SPL2, and the data plumbing behind detection engineering in a large health system. Agentic AI systems. Agents that use tools under human oversight, how they are tested, and the runtimes that host them. aitester-bdd has an agent write real end-to-end tests by using the app it tests. Bayesian and statistical modeling. I write open books that teach the methods by building them: Bayesian Modeling with PyMC and Advanced Methods in SciPy and Statsmodels. Software that runs on your own devices. Through bayeslearner LLC I make Screenbay, an iPad app for the Screen Sharing built into macOS. I also run a small private cloud at home. Tools for writing and thinking. Obsidian plugins for review comments (Sidemark), pen input (Handwriting) and print-quality rendering (Typeset). Get in touch # Email hello@bayeslearner.org. I read every message.\n","externalUrl":null,"permalink":"/","section":"Kun Deng","summary":"","title":"Kun Deng","type":"page"},{"content":"Most of these are open source. Each says where its code or documentation lives.\nApps # Screenbay # An iPad app for the Screen Sharing built into macOS. It shows your Mac\u0026rsquo;s desktop on the iPad at full Retina resolution, with sound, over your own network. No account, relay server or cloud service is involved. Made by bayeslearner LLC. Overview\nremotex # Remote desktop in a browser for the servers each platform already has: Windows Remote Desktop, macOS Screen Sharing in both of its modes, and Linux through a VNC server for Wayland compositors. A small gateway you run next to your machines. Code\nDeveloper tools # appbay # A Docker-native PaaS control plane with Docker Compose as the application model. More · Code\naitester-bdd # End-to-end testing for any web app: describe the test in English, and an agent writes a runnable Robot Framework test by actually using the app. Docs · PyPI\nplugkit # A plugin kernel for Python in which every registration returns its undo and a fiber owns it. Code\nSplunk SPL2 wiki # A reference library for Splunk\u0026rsquo;s SPL2, with a scenario cookbook and a developer tutorial. Code\nObsidian plugins # Sidemark # Google Docs-style comments and suggested edits for Obsidian, stored beside each note in the open MRSF format so the Markdown stays clean. Plugin · Code\nHandwriting # Handwrite, highlight, erase and lasso directly on notes and PDFs; the notes stay Markdown. Code\nTypeset # Typora-style document rendering and PDF export inside Obsidian. More\nOpen books # Bayesian Modeling with PyMC: From First Principles to Practice: Bayesian fundamentals, hierarchical models, Gaussian processes, nonparametric Bayes and causal inference, in 11 chapters. Advanced Methods in SciPy and Statsmodels: optimization, Monte Carlo, likelihood inference, time series, multivariate methods and causal inference, with from-scratch implementations, in 23 chapters. ","externalUrl":null,"permalink":"/projects/","section":"Projects","summary":"","title":"Projects","type":"projects"},{"content":"Data architect · security data engineer · machine-learning researcher · Ann Arbor, MI · hello@bayeslearner.org\nEmployment # Michigan Medicine (University of Michigan Health System) # Splunk Data Architect and Data Scientist · Ann Arbor, MI · 10/2013 – present\nSince 2026: security data engineering in Information Assurance, building the data pipelines and searches behind detection engineering. Splunk architecture for a large health system: ingestion, data governance, dashboards and apps. Time-series analysis of machine and health metrics with Splunk, Tableau and Python. Adjunct Lecturer in the University of Michigan\u0026rsquo;s Master\u0026rsquo;s Degree Program in Data Science. bayeslearner LLC # Founder · Michigan\nBuilds and sells focused software that keeps data on the user\u0026rsquo;s own devices, starting with Screenbay for iPad. University of Michigan # Postdoctoral Research Fellow · Ann Arbor, MI · 09/2010 – 09/2013\nAdvanced clinical-trial recruitment through active and budgeted learning algorithms. Data transformation and analysis in R and SAS. Taught Programming and Numerical Methods in Statistics (Stats 607) in Python. Arizona State University # Postdoctoral Research Fellow · Tempe, AZ · 09/2009 – 09/2010\nLed R\u0026amp;D for computer-aided diagnosis of pulmonary embolism. Machine-learning methods built on cascaded AdaBoost. University of Nebraska # PhD Research Assistant · Lincoln, NE · 09/2004 – 09/2009\nR\u0026amp;D of machine-learning algorithms in MATLAB, C++ and Python. Taught the undergraduate MATLAB programming course. Education # Ph.D., Computer Science, University of Nebraska, 2009 M.Sc., Computer Science, National University of Singapore, 2003 M.Eng., Electrical Engineering, Beijing Normal University, 2001 Skills # Languages: Python, R, TypeScript, SQL, SPL and SPL2, PowerShell, MATLAB, Java Data and security pipelines: Splunk, Cribl, NiFi, Spark, Kestra, n8n AI and machine learning: agentic systems and LLM tooling, Bayesian modeling (PyMC), active and budgeted learning, predictive modeling Infrastructure: Docker and Compose, Proxmox, Nomad, OpenTofu, Ansible Visualization: Splunk dashboards, Tableau, Power BI Certifications: Splunk Certified Architect, Tableau Certified Professional, RapidMiner Certified Professional Selected publications # Deng, Zheng, Bourke, Scott, Masciale. \u0026ldquo;New Algorithms for Budgeted Learning.\u0026rdquo; Machine Learning 90(1), 2013. Deng, Pineau, Murphy. \u0026ldquo;Active Learning for Personalizing Treatment.\u0026rdquo; IEEE ADPRL, 2011. doi:10.1109/ADPRL.2011.5967348 Wu, Deng, Liang. \u0026ldquo;Machine Learning Based Automatic Detection of Pulmonary Trunk.\u0026rdquo; Proc. SPIE 7963, 2011. Zheng, Scott, Deng. \u0026ldquo;Active Learning from Multiple Noisy Labelers with Varied Costs.\u0026rdquo; IEEE ICDM, 2010. Bourke, Deng, Scott, Schapire, Vinodchandran. \u0026ldquo;On Reoptimizing Multi-Class Classifiers.\u0026rdquo; Machine Learning 71(2–3), 2008. Deng, Bourke, Scott, Sunderman, Zheng. \u0026ldquo;Bandit-Based Algorithms for Budgeted Learning.\u0026rdquo; IEEE ICDM, 2007. Culver, Deng, Scott. \u0026ldquo;Active Learning to Maximize Area under the ROC Curve.\u0026rdquo; IEEE ICDM, 2006. ","externalUrl":null,"permalink":"/resume/","section":"Kun Deng","summary":"","title":"Resume","type":"page"},{"content":"","externalUrl":null,"permalink":"/series/","section":"Series","summary":"","title":"Series","type":"series"},{"content":"","externalUrl":null,"permalink":"/tags/","section":"Tags","summary":"","title":"Tags","type":"tags"},{"content":"","externalUrl":null,"permalink":"/posts/","section":"Writing","summary":"","title":"Writing","type":"posts"}]