Jonathan Scarpelli
Machine Learning Engineer, Machine Learning Researcher, Data Scientist, and Full-Stack Developer
I build developer tools and study how research claims hold up under evaluation: coding-agent completion, financial models and cross-study transfer.
Current research
Research & working papers
These papers are working manuscripts, not peer reviewed. AI tools were used throughout the research workflows. Each project summary describes individual and collaborator contributions and the limits of the evidence.
All papers & contributionsCommandem
Working manuscript · October 2026 · Not peer reviewed
A retrospective audit of what completion markers, passing checks and recorded agent behavior actually establish.
Quantem
Working manuscript · October 2026 · Not peer reviewed
An audit of timing, transaction costs and graph evaluation in relational stock prediction, separating controlled evidence from recovered historical outputs and conditional replay.
Oralbiome
Collaborative working manuscript · Not peer reviewed
A retrospective audit of oral-microbiome diabetes-status models, focused on cohort eligibility, leakage, calibration and cross-study transfer.
Earlier product work
Oasys
Built an AI-powered real estate platform with virtual assistants for property transactions.
Blackout Lighting Console
Developed and managed an iPad lighting console for controlling fixtures in professional productions.
Jotto
Built a text and chat platform connecting government agencies and nonprofits with their communities.
STAR
Built a student-outcomes analytics platform for the University of Hawaii system.
Products · Scarcrux
Less switching.
More doing.
Scarcrux is where I build tools for everyday workflows: coordinating coding agents, finding and moving files, and running streaming communities.
Explore all productsCommandem
Private betaBring projects and coding agents into one workspace. Plan across repositories, dispatch work, and review changes with verification evidence.
- Plan across repos
- Dispatch agents
- Review evidence
Hello World, I'm Jonathan
My work focuses on causal discovery, graph neural networks, and applied machine learning across finance and biomedical research. I bring a product-development background to that work, having shipped web and mobile apps and cofounded a real estate startup. I'm interested in research collaborations and opportunities to build useful ML systems.
Outside of training models, I like to read about everything, play video games, watch movies, visit new places, eat my way through NYC's food scene, pamper my 4 cats and a dog, and tend to my indoor and outdoor gardens. Always open to interesting opportunities — feel free to drop me a line!
Send me a message