About
I'm a software engineer at NICE CXone Knowledge with 5 years of building for the web. I came to software from bioscience, with a chemistry degree and lab quality control. I started in technical support, moved into QA automation, found my home in front end, and now work across the backend too, so I think about the whole product lifecycle: what users struggle with, how it should be built, and how it holds up once it ships.
I work AI-first. Claude Code, Cursor and custom MCP tooling are part of how I plan, build, review and debug, and I'm deliberate about where they help and where human judgment still wins. My MCP tools and skills help me find the root cause of bugs faster, and I scope work with product and design so it ships.
I also like leveling up the people around me: I train and mentor teammates through pairing, setup guidance and docs. Away from the keyboard you'll find me playing pickleball, traveling, hunting for the next great meal, or at a concert.
Experience
Jan 2019 – Present
NICE CXone Knowledge (formerly MindTouch)
An enterprise knowledge-management SaaS that grew from a wiki powering support sites into the knowledge layer for omnichannel experiences like chatbots and AI.
Software Engineer
Apr 2021 – Present · promoted from QA
- Own features from requirements through release, across the front end in React, TypeScript and Vue.js and, now, the backend in C# on AWS, working closely with product, design and QA.
- Delivered Filter Groups for UnifiedKnowledgeService, which feeds filtering knowledge (properties, metadata, tags and other fields) to RAG-powered LLM experiences. Built with agentic workflows at roughly 3x speed, and highlighted among my team.
- Built MCP tools and skills that track down bug root causes: they open internal endpoints, sign in to sites, query our log tools and reproduce issues on test sites.
- Train and mentor teammates on AI-assisted workflows through pairing, setup guidance and documentation.
- React
- TypeScript
- Vue.js
- C#
- AWS
QA Engineer
Nov 2019 – Apr 2021
Embedded QA engineer automating tests for new features with Selenium, C#, JavaScript, Jenkins, Codeship, Docker and AWS. I learned to see the product the way it breaks, and that instinct still shapes how I build and review code.
Support Agent, then Technical Customer Success Agent
Jan 2019 – Nov 2019
On the front line of user pain points, the foundation for how I prioritize and ship.
Dec 2017 – Jan 2019
MadCap Software
Technical Support Analyst
Dec 2017 – Jan 2019
2014 – 2018
Earlier experience
Web Developer · Hoop-ball.com (remote)
Feb 2018 – Aug 2018
Full Stack Developer · Origin Code Academy
Jun 2017 – Nov 2017
Quality Control Analyst · Stemedica Cell Technologies
Dec 2014 – May 2017
Quality Control Chemist · J & D Laboratories
Jul 2014 – Nov 2014
Skills
- Front end
- React · TypeScript · Vue.js
- Back end & cloud
- C# · AWS
- Testing & CI/CD
- Vitest / Jest · Jenkins · Docker
- Observability
- Datadog · Sentry
- AI tooling
- Claude Code · Cursor · GitHub Copilot · MCP servers & custom skills
Projects
Chatchat.matthew-tran.com ↗ (opens in a new tab)
A ChatGPT-style chat app I built end to end. Guests can jump right in, Google sign-in saves your chat history, and every model call is logged so I can compare models. It's a live proof that I practice the AI-first engineering I talk about.
Built with: Next.js · React · TypeScript · Tailwind · Baseten (LLM inference) · Supabase (Postgres + auth) · Vercel
View source ↗ (opens in a new tab)Password: . It runs on a $20 monthly spend cap, so if it isn't responding the cap may have been hit. It also relies on free models, which can be rate limited or exhausted at times. If it doesn't work, try again later.
Matthew Tran Shopshop.matthew-tran.com ↗ (opens in a new tab)
I'm a passionate custom t-shirt maker, and every design is something I'd actually wear: minimal, personalized tees for moms and dads, printed to order. The shop is about the things that matter most to me right now: being a dad, and being an active member of my local community, including a Mira Mesa collection.
Unpaywallunpaywall.matthew-tran.com ↗ (opens in a new tab)
A small utility that turns a pasted link into a clean, readable article. It works through a fallback chain, trying Googlebot-style access first, then archive.ph and the Wayback Machine, and is upfront when a site can't be opened. It's a good example of how I like to build: a narrow problem I ran into myself, a simple interface, and honest limits.
Built with: Next.js (App Router) · TypeScript · Vercel · SSRF guard and locked-down CSP
View source ↗ (opens in a new tab)Baby Dashboardhuckleberry.matthew-tran.com ↗ (opens in a new tab)
A dashboard for the daily rhythm of life with a baby: sleep, feeds, diapers, pumping and growth in charts and a day-by-day timeline, plus an "ask a question about your data" box for quick answers like how many diapers today. Turning a pile of log entries into something a tired parent can read at a glance.
Built with: ECharts · unofficial Huckleberry API (refreshed every 30 minutes) · WHO growth standards · rule-based Q&A that runs in the browser
Say hello
I'm always happy to talk about engineering, AI workflows, or what you're building.
Start a conversationor email matthewhaotran@gmail.com




