Selected work.

Client delivery, products we build, and technical demonstrations. Each account identifies our role and the stage of the work. Client details stay private.

Subcontract delivery

Hospital application streaming.

Pilot documentation delivered

A hospital assessing a move from Citrix to AWS WorkSpaces Applications needed to establish how its applications would be packaged, delivered, and tested.

0pon’s founder prepared a four-application pilot under subcontract, built repeatable image-installation scripts, and delivered application discovery, architecture, configuration, and pilot-result documentation.

What the pilot established: installation and streamed launch paths. Clinical workflow completion and a production migration were outside the demonstrated result.

Practice integration

A dental practice moving off paper.

In deployment

A paper-based dental practice is moving onto an open-source practice-management system, with 0pon as the integrator. The design hosts the system on a machine at the clinic, with backup, recovery, and access procedures included in the deployment.

The practice owns its licenses and data. Integration is in progress.

Practice workflow and payment audits

0pon Care product · Synthetic-data demonstration

Separately, 0pon Care brings referrals onto a staff review board and compares claim payments with contracted fee schedules. The working demonstration uses a synthetic practice with no real patient data, showing referral review, claim comparisons, and practice-system integration using sample records.

Explore the dental software

Owner-built product

Commuter.

Invite-only

Job searching spreads the same information across listings, resumes, and application forms. Commuter ranks opportunities against a person’s background, prepares application documents, and keeps application tracking in one place.

A browser extension fills application forms for the person to review and submit. The captioned walkthrough uses sample data to show the product in use.

Watch the Commuter demo and read the docs

Internal proof of concept

Model tracking and promotion.

Sandbox demonstration · October 2026

Connected Red Hat OpenShift AI workbenches and training pipelines to MLflow experiment tracking and model registration in Red Hat Developer Sandbox. The question: can a training run record its result and use it to decide whether to replace the current model?

Two completed runs, two decisions

  • A lower-scoring candidate was registered; the current model was retained.
  • A better-scoring candidate was registered and promoted to the current model.

This was an independent technical demonstration. The assessment identified authentication, reproducibility, and promotion-history gaps to resolve before production use.

Research

n0de: research into access-control guarantees.

n0de is access-control research for multi-level data: one identity holding many compartments at once. Specific properties of its formal model are machine-checked in Lean 4. The model covers no read-up and no write-down across level, compartment, and need-to-know; non-interference between compartments; and what a user can observe after access is revoked. These guarantees apply to the model and its assumptions.

The research compares Cedar policy decisions with the model and checks the correspondence of a PostgreSQL row-level-security predicate. This does not establish that a complete deployed system cannot leak data. An SBIR proposal building on this work is under review.

This page describes the research; it does not publish n0de’s proof files or verification runs. Separate threshold-disclosure research has public source code. It is a different project, not evidence for n0de’s access-control properties.

Have a problem like these?

Send a short note