Desktop / local AI
CareerOS Local
A private desktop workspace that turns verified career records into focused searches, stronger applications and a practical agenda for the next step.
- Python
- React
- Rust
Ejupi Labs / Open-source product archive
Seven open-source products, one maintained archive.
Public work spanning local AI, machine learning, automation, network tools, numerical methods, operations and consent-based support. Each record provides context, then links to the product, source and automated checks.
Product archive
Products in the public record.
Each entry states what the product is for, identifies its core technology and links to its source and automated checks.
-
-
Machine learning
ELIZA Lab
An inspectable open-set machine-learning lab for training, calibration and abstention, backed by frozen inputs and reproducible artifacts.
- Rust
- Open-set ML
- Calibration
-
Controlled automation
Djenis
AiAgent A permission-aware agent for Windows and browser tasks, using scoped capabilities, explicit allowlists and fresh-state verification.
- Python
- Browser control
- Windows
-
Network protocol
DIG
A bounded Gopher client with a CLI, a local web inspector and a same-origin gateway that keeps selectors, routing and response limits visible.
- Node.js
- Gopher
- Network policy
-
Numerical methods
IntegraDraw
A visual workbench for comparing midpoint and trapezoidal integration against a shared Simpson reference.
- Java
- TypeScript
- Numerics
-
Operations workspace
VECTOR
A self-hosted placement operations platform with scoped roles, persistent records, completion controls and an append-only audit trail.
- Node.js
- SQLite
- Operations
-
Consent-based remote support
JDoor Assist
A remote-support project built around explicit invitation, observable sessions, immediate revocation and gated distribution.
- Java
- Session consent
- Remote support
Editorial standard
How to read the archive.
The index is intentionally concise. Detail remains close at hand without turning the index into a substitute for the product or code.
Product context
Purpose and operating boundaries appear before implementation detail.
Direct routes
Product and Source stay visible and consistently placed in every record.
Inspectable evidence
Limits, tests and release status remain documented where they can be verified.