Khoa Nguyen
BS in Computer Science, MGA.edu - 4.0 GPA
BS in Bioinformatics, ASU.edu - 3.8 GPA
Professional Summary
Senior Software Engineer with 10+ years of experience building and modernizing backend and platform systems using PHP and Laravel.
Read more
Strong background in legacy system modernization, performance optimization, infrastructure reliability, and developer productivity. Hands-on experience integrating AI-assisted tooling into real-world development workflows while maintaining production-grade quality and security standards.
Skills
Projects
Bills Helper Personal-finance manager with an AI-queryable read-only API
Tracks accounts, bills, income, transactions and 0% APR promotional balances, then forecasts balances forward so a shortfall shows up before it lands.
- Laravel 13
- PHP 8.3+
- Livewire 3 + Volt
- Flux UI
- Pest 4
- Playwright
- GitHub Actions
- Modelled asset, revolving and liability accounts under one ledger, where debt accounts invert debit/credit and read as "owed" app-wide, with recurrence and payment logic isolated into reusable concerns.
- Built 0% APR promotional-balance tracking: expiry awareness plus the payment required to clear a balance before deferred interest is charged retroactively.
- Forward-looking per-account balance forecasting that flags projected shortfalls ahead of time.
- Shipped a read-only MCP server and an in-app AI assistant as thin adapters over one shared query layer, so both surfaces answer from identical logic.
- Both AI surfaces are read-only and answer only for the authenticated user: the identity comes from the token, never from the request's own input.
- Public demo on an isolated deployment seeded with synthetic data that resets daily.
- Pest 4 unit and feature suite on in-memory SQLite plus a Playwright browser suite, gated in CI alongside linting, commit conventions and secret scanning.
- Runs in production as my own day-to-day finance tool, so every feature is exercised against real use before it ships.
Materia Claude-powered pull-request review, packaged as a portable GitHub Action
Sockets an AI code review into any repository: one workflow file and an optional config file, with nothing added to the project's own dependencies. Wraps the official Claude Code action behind a gate, a config layer and a tuned prompt library.
- GitHub Actions
- Claude Code Action
- Node.js
- JavaScript
- Node test runner
- Packaged the whole review as a versioned composite action, so adopting it costs one workflow file plus an optional config file and adds nothing to the consuming project's dependency tree.
- A deterministic gate decides whether to review before any model runs, reading draft state, target branch, changed paths, labels and diff size, so a pull request that should be skipped costs nothing.
- Fork pull requests never see secrets, because the action uses the plain pull-request trigger only and skips cleanly instead of escalating its own permissions.
- The reviewer is read-only by construction: Claude gets file-read and search tools only, with no shell, no writes and no push, and the summary comment is posted by the action's own trusted step rather than by the model.
- A second mode turns the same diff into a manual-QA guide, posted under its own comment so it never collides with the review.
- Every deterministic step is a plain module covered by unit tests: context detection, config resolution, the gate, prompt assembly. The action manifest, its workflows and secret scanning are all linted in CI by pinned, checksum-verified tools.
- Built an eval harness so review quality is a measured number over a labelled corpus rather than an impression; it runs today on synthetic cases while real ones are labelled.
- Review memory: a finding a human dismissed is recorded and not raised again on later runs, so repeat reviews stop re-litigating settled points.
- Reviews its own pull requests, so every prompt and guardrail change is exercised on real diffs before it ships.
Case Study
Dealer Auto-Order Pipeline: how the nightly orders from roughly 3,000 storefronts were accepted synchronously, fulfilled on a queue across a dozen vendor integrations, and recovered after failure without ever double-sending a purchase order.
Academic Experience
BS in Computer Science, Middle Georgia State University, 4.0 GPA
Concentration in Software Development
- Graduated with Honors; completed 106% of required credits
- Senior Capstone, Internship, and 15+ upper-level IT courses including Python, Java, C++, InfoSec, and Web Dev
BS in Bioinformatics, Arizona State University, 3.8 GPA
- Completed advanced coursework in Genetics, Evolution, Ecology, and Computational Biology
- Early focus in life sciences before pivoting to tech; cross-disciplinary foundation