Hoeurb VuthiDeveloper · Phnom Penh, Cambodia
Observe — Study how the system behaves 01 / 05

Software, from
the inside out.

I’m Hoeurb Vuthi, a developer and MIS student. I build web applications, explore local AI, and learn by taking systems apart.

Explore my work Get in touch

C++ foundations. Web interfaces. AI experiments.

Selected work

A closer look at the systems I’m building, from browser to background process.

Project details: Agentic AI Assistant

Agentic AI Assistant

A sandboxed, autonomous local AI assistant operating 24/7 with Docker container isolation, prompt injection defense, and automated task execution.

Architecture schematicIllustrative architecture — details pending
Agentic AI Assistant architecture schematicA request passes through the prompt injection guard to a task runner. Recurring tasks enter through the scheduler. The runner dispatches work into a Docker sandbox; captured output and exit state feed the status dashboard. Internal stages and fields are illustrative.Request inputTask / contextExecution scopePrompt guardScreen task inputPrompt injection defenseSchedulerRecurring tasks24/7 schedulingTask runnerDispatch guarded tasksScheduled + on-demandDocker sandboxIsolated local executionRun task / capture resultRun outputOutput / exit stateExecution resultStatus dashboardLive task stateResult / activity historyAgentic AI Assistant architecture schematicA request passes through the prompt injection guard to a task runner. Recurring tasks enter through the scheduler. The runner dispatches work into a Docker sandbox; captured output and exit state feed the status dashboard. Internal stages and fields are illustrative.Request inputTask / contextPrompt guardScreen task inputSchedulerRecurring tasksTask runnerDispatch guarded tasksDocker sandboxIsolated local executionRun outputOutput / exit stateStatus dashboardLive task state
View Agentic AI Assistant
Project details: AI Models Finder

AI Models Finder

A website that helps nontechnical people understand, find, and choose AI models for their needs.

Architecture schematicIllustrative architecture — details pending
AI Models Finder architecture schematicMultiple benchmark sources provide verified scores across several metrics. A recommendation algorithm matches those metrics to user needs and requirements, producing model rankings for comparison and selection. This feature-based architecture is illustrative.BenchmarksMultiple sourcesModel benchmark resultsVerified scoresVerified benchmarksModel / score / sourceMetric coverageMultiple measuresUse-case relevanceUser needsUse cases / requirementsRecommendationMatching algorithmNeeds + benchmark scoresModel rankingsRanked model matchesBenchmark-based scoresCompare modelsReview metrics / fitChoose a modelAI Models Finder architecture schematicMultiple benchmark sources provide verified scores across several metrics. A recommendation algorithm matches those metrics to user needs and requirements, producing model rankings for comparison and selection. This feature-based architecture is illustrative.BenchmarksMultiple sourcesVerified scoresVerified benchmarksMetric coverageMultiple measuresUser needsUse cases / requirementsRecommendationMatching algorithmModel rankingsRanked model matchesCompare modelsReview metrics / fit
View AI Models Finder
Project details: Task Automation Bot

Task Automation Bot

CLI daemon for automated file classification, scheduled system backups, and recurring background operations with SQLite rollback logs.

Architecture schematicIllustrative architecture — details pending
Task Automation Bot architecture schematicA file watcher feeds classification rules, while a cron-style schedule supplies recurring jobs to the Typer CLI worker. Custom plugins extend the worker. File organization, backups and email jobs produce SQLite rollback records and audit output. Internal stages and storage relationships are illustrative.File watcherWatch file changesDiscover incoming filesFile rulesMatch / classifyExample: *.pdf → docsScheduleCron-style jobsRecurring backups / emailCustom pluginsAutomation modulesExtend task behaviorCLI workerTyper command entryPlan / dispatch tasksJob executionOrganize / backup / emailRun selected operationSQLite logTask / change recordsRollback supportAudit outputTime / task / resultReview run historyTask Automation Bot architecture schematicA file watcher feeds classification rules, while a cron-style schedule supplies recurring jobs to the Typer CLI worker. Custom plugins extend the worker. File organization, backups and email jobs produce SQLite rollback records and audit output. Internal stages and storage relationships are illustrative.File watcherWatch file changesFile rulesMatch / classifyScheduleCron-style jobsCustom pluginsAutomation modulesCLI workerTyper command entryJob executionOrganize / backup / emailSQLite logTask / change recordsAudit outputTime / task / result
View Task Automation Bot
Browse all 4 projects

Curiosity is
the starting point.

I’m a Management Information Systems student at SETEC Institute in Phnom Penh, working toward becoming a full-time software developer.

I started with C++ and Linux tooling, then found my way to modern web development. Today I build with Astro and TypeScript, experiment with local AI and automation, and keep sharpening my problem-solving skills.

Straightforward code. Fast interfaces. A little more understanding with every project.

Resume status

Web & interfaces

Astro · TypeScript · HTML · CSS · Tailwind CSS

Systems & tools

C++ · Python · Linux · Bash · Git · Docker

AI & automation

Local LLMs · Ollama · API integration · Task workflows

Visual practice

Photoshop · Illustrator · Lightroom · Layout design

Let’s build something useful.

Have a project in mind, a question about my work, or an interesting problem? I’d like to hear it.

GitHub (opens a new tab)LinkedIn (opens a new tab)

Delivered through Web3Forms.

Inside the project

A public demo or repository link hasn’t been supplied for this record.

Portfolio assistant

Ask about Hoeurb’s work and experience.

AssistantWhat would you like to know about Hoeurb’s projects, skills, or background?