KEVIN NGUYEN -- ENGINEER, TINKERER, NIGHT OWL
I'm Kevin. By day I write the software behind a smart TV home screen. Late at night I'm in the workshop: building tools for my own AI agents, turning what I learn into notes I can find again, and sometimes queueing ranked TFT instead.
KEVIN NGUYEN -- ENGINEER, TINKERER, NIGHT OWL
I'm Kevin. By day I write the software behind a smart TV home screen. Late at night I'm in the workshop: building tools for my own AI agents, turning what I learn into notes I can find again, and sometimes queueing ranked TFT instead.
01 — Human
I can't leave a problem alone. Most of what I do is some version of hunting for the fix: debugging embedded C++ at work, tinkering with AI tooling at home long after I should be asleep, or working out why I just lost a TFT lobby. I learn by building something, breaking it, and staying until it clicks. The part I like most comes after, when the person I was helping is unstuck too.
cats
ranked
also playing
otherwise
POSITION OVER NAMES
Folder structure is the API.
Position-addressed memory beats named imports. Names rot when you rename things; positions don't. The whole site is built on this idea — your folder paths ARE your routing, your taxonomy, your contract. ICM makes this explicit.
02 — Workspace
Position-addressed, harness-agnostic -- the knowledge architecture I build for teams, running my own life. Browse it. README.md is open for you.
03 — Capabilities
Multi-step LLM pipelines with tool use, triage cascades, and human-in-the-loop gates -- built to run unattended and fail loudly.
Position-addressed knowledge bases, retrieval indexes, and CLAUDE.md/CONTEXT.md routing that any LLM harness can navigate cold.
Hard verification gates, automated test generation, requirement-to-test mapping, and refactoring pipelines that cannot ship silent breakage.
Took a production C++ team from 0% to daily AI use: workshops, live demos, custom skills, and SOPs that survive after the champion leaves the room.
Custom agents, skills, hooks, and MCP integrations; Python pipeline tooling that turns one-off AI wins into repeatable team infrastructure.
Five years of production C++/Qt on John Deere displays, 100+ defects resolved, now smart TV UI at VIZIO -- the engineering rigor underneath the AI work.
04 — Pipeline
a video link or a web clip
skip / light / deep routing
structured claims, timestamp anchors
every claim needs an anchor, three spot-checked
cross-linked synthesis layer
humans + agents query one index
a video link or a web clip
skip / light / deep routing
structured claims, timestamp anchors
every claim needs an anchor, three spot-checked
cross-linked synthesis layer
humans + agents query one index
05 — Runbook
06 — Changelog
Every role since 2020, newest first. Open an entry for the full diff.
August 2026 – Present/Dallas, TX
The home screen and menu on VIZIO smart TVs: embedded Linux, C++, and a custom declarative UI framework.
Software Engineer II on the team that builds the home screen and menu experience for VIZIO smart TVs. Works across an embedded Linux platform in C++, a custom declarative UI framework, and Python tooling. Joined in August 2026 after five years on John Deere guidance displays, and brings the same evidence-first debugging and test discipline to consumer TV software.
C++, Embedded Linux, Declarative UI, Python
VIZIOJune 2024 – July 2026/Austin, TX
Shipped autosteer features in C++/Qt and built the team's AI workflow: a knowledge system, agentic skills, and the rollout from zero to daily use.
Developed features for AutoPath: designing, implementing, and testing the supported product lines (Boundaries, Rows, Zones, Pro) to ensure high quality. Led a team refactor of the most-modified service class in the AutoPath codebase, automating it with a refactor tool that combines Python scripting and agentic tooling in a 7-layer pipeline, following the company SDK and practices from Working Effectively with Legacy Code (Michael Feathers). Built a documentation system serving as the team's living PRD, onboarding hub, and AI-consumable knowledge base, covering 4 product lines (Boundaries, Rows, Zones, Pro), 30+ vehicle models, and 26 features. Created 6 specialized AI skills and scripts for automated test generation, requirement extraction, test coverage analysis, test-to-requirement mapping, refactoring with safety nets, and LLM architecture diagramming. Drove GitHub Copilot adoption across the team from 0% to daily use through presentations and demos on agentic engineering, CLI workflows, prompting techniques, and AI-assisted refactoring.
C++14, Qt5, Embedded Linux, Google Test, CMake, Python, MkDocs
AutoPathFebruary 2023 – June 2024/Austin, TX
Deep-debug specialist on Gen4 displays: crashes and threading defects traced to root cause with GDB, valgrind, and ASAN. Guidance Department Star Award.
Deep-debug specialist resolving 38 major and 60+ minor defects across Gen4OS 4060 & 4600 displays. Diagnosed SIGSEGV/SIGABRT crashes and threading defects (race conditions, deadlocks, priority inversion) using GDB, valgrind, and ASAN. Executed SiL (Software-in-the-Loop) and on-vehicle verification testing with tractors and combines; used Pumba (Java-based vehicle simulator with HTTP REST API) to drive automated system-level tests without physical hardware. Debugged vehicle defects using CAN data analysis tooling — ATPerformanceTool for live CAN bus monitoring and CAN Sniff for log-based vehicle signal analysis with MATLAB Simulink visualization. Received one of six Guidance Department Star Award for resolving the team's most complex escalated defects, standing in as team lead during manager absence, and delivering senior-level technical updates directly to department leadership.
C++, Qt, GDB, Valgrind, ASAN, Google Test, Froglogic Squish, CAN Protocol, MATLAB Simulink, Pumba
tag: B.S. Computer Engineering -- Iowa State University, December 2022
July 2021 – December 2022/Ames, IA
Embedded software testing while a full-time student at Iowa State. Taught myself C++ and Qt in the first six months.
Conducted embedded software testing while enrolled full-time at Iowa State. Resolved 24 confirmed defects using Google Test and Froglogic Squish. Refactored AutoTrac PieState logic into an independent component. Self-taught C++ and Qt through Udemy courses within 6 months of joining.
C++, CMake, Qt, CAN Protocol, Ubuntu Linux
January 2022 – December 2022/Ames, IA
Senior design with a medical device startup: led the front end of an inventory and shipment tracker that replaced Excel.
Senior design project with a medical device startup that needed a custom inventory site. Led front-end development on a 5-person student team, building an internal inventory and shipment tracking web app that replaced manual Excel workflows with a full-stack React application.
React, JavaScript, CSS, AWS Amplify, AWS DataStore, AWS S3
February 2020 – July 2021/Ames, IA
Primary Apple technician: Jamf Pro deployments, Python device-management scripts, and training for new part-time techs.
Primary Apple technician deploying Jamf Pro and writing custom Python scripts for device management. Administered software licenses, performed equipment installations, OS imaging, and hardware/software troubleshooting for staff workstations and portable devices. Coordinated with full-time IT staff remotely for advanced troubleshooting. Led new part-time technicians on help desk tasks including hardware/software installations, macOS system administration, and inventory management and deliveries.
Python, Jamf Pro, macOS, Windows, Help Desk
07 — Contact
Stuck on something, building with agents, or want to trade TFT comps and basketball takes? Say hi. The resume is here if you want the formal version, and my agent can answer most questions about me.
or email kevtrinhnguyen@gmail.com
I take engineering teams from "AI is a demo" to "AI is how we work": multi-agent pipelines with eval gates, knowledge systems both humans and agents can navigate, and the enablement that makes it stick.
Now: Software Engineer II at VIZIO, on the smart TV home screen and menu (since August 2026). Dallas, TX.
Evidence: a 7-layer agentic refactor pipeline on shipped C++ at John Deere, a team knowledge hub covering 30+ vehicle models, 0% -> daily Copilot adoption, and ideaverse-os (open source).
Foundation: five years of production embedded C++/Qt on John Deere guidance displays (2021-2026).
Position-addressed: folder = type, prefix = priority. New work earns a folder slot; old work doesn't get to take up visual real estate. Rules and identity for all AI agents live in AGENTS.md (canonical); CLAUDE.md and copilot-instructions.md are 1-line shims pointing back to it.