I create systems that can think, products you can hold, and soundtracks worth hearing.
I'm an applied computer scientist. At Treobytes I shipped a production AI agent system on the Claude API that cut a four-day document process to five hours. I build hardware end to end, from firmware and software to CAD and manufacturing, and I compose award-winning music for film and games.
Production Claude API agents, RAG, LLM-as-judge review, internal React platforms.
Hardware & CAD
Circuit, USB-HID firmware, CAD, and a buildable MIDI controller kit produced at scale.
Audio & Music
Award-winning scores and adaptive game-audio systems across 31 credits.
Teaching
University sound-design professor, teaching game audio from DAW fundamentals to adaptive systems.
Selected Work
Things I've shipped
Filter by discipline. Each project in short: what it is, my role, and what came of it.
AI & Engineering2025 · Treobytes
Claude AI Skill Tree
A production AI agent system on the Claude API that generates compliant, submission-ready proposals and grant documents from live data via RAG. I designed and shipped it solo on a three-layer skill architecture that non-technical users operate.
Cut team turnaround from four days to five hours, with an automated LLM-as-judge review flagging policy and brand-guideline violations before anything reaches a client.
Claude APIRAGLLM-as-judgePythonJSON schema
Hardware & CADAI-assisted build2026 · Independent
CADence
A browser 3D modeler that makes precise, professional-grade CAD as approachable as a toy. Every shape is stored as an editable recipe of exact parameters, which is what makes the freeform-to-precise switch possible.
Solid modeling with exact dimensions and clean STEP export, plus the move beginner and pro tools both miss, switching between freeform sculpting and precise, editable parametric design on the fly.
A buildable MIDI keyboard kit that teaches students to wire real circuitry. I designed the circuit, the USB-HID firmware, and the laser-cut enclosure end to end, safe and simple enough for a classroom.
Students assemble it themselves, placing real components onto a breadboard over an accurate schematic overlay. Plug-and-play with no drivers, compatible with any standard DAW.
An audio system that reshapes a game's musical score in real time. Composers write the stems by hand, and the engine rearranges and warps them live with player movement and action. Demoed end to end in Night Shift, my thesis game, with voice acting, UI art, and logo by Emanuel Nation.
Showcased at the Game Developers Conference, San Francisco 2024, and shipped as the audio engine of an indie studio's game.
An internal React platform I built from the ground up, pairing a daily-use facilitator app with a management admin console.
Adopted org-wide as the central hub for daily operations and oversight, including a seating-chart generator that pairs exhaustive search with a greedy heuristic to solve constraints in seconds.
ReactNode.jsExpressPostgreSQL
Hardware & CAD2021 · Research
KUKA Robotic 3D-Printing Research
Sole programmer on my university's first large-scale robotic 3D-printing team, turning generative algorithms into sculptures printed by a KUKA robotic arm.
I wrote Python generative pipelines in Rhino and Grasshopper that produced sculptural geometry and fabrication-ready exports for a custom print head.
A data sonification that turns the rise and fall of Google search interest into music. Three search terms scroll down a piano roll, each voiced by its own instrument, so a dataset becomes something you hear as much as read.
I built it in p5.js, mapping search-interest values to notes in C major and playing them live on string, piano, and violin as the timeline scrolls.
An original soundtrack I wrote, orchestrated, mixed, and mastered, recorded with live session musicians and featured in IndieWrap's 2023 short-film collection.
Silver, Best Composer, Tokyo Film Awards 2022; Best Original Music nominee, New Creators Film Awards 2022.
A streetwear brand with cinematic drops. I scored several of its films and campaign spots and built the sweetcream.co storefront around the brand's identity.
Music and web design carried the Fall 2022 launch; the archive preserves the original site, lookbook, and store page for page.
Web designOriginal scoreCompositionSound designFoley / SFXMixingHTML / CSS
I contributed front-end development to a shipped English and Spanish translator tool for AltaMed, a healthcare provider, so bilingual patients and providers could communicate clearly. Accessibility and bilingual content stayed the priority throughout.
Shipped to production, and I built WebFlow demos for a broader healthcare-site overhaul.
A 3D digital prototype built for contemporary artist WangShui. I modeled their sculpture concepts in Maya, then recreated the gallery in Unity as a walkable, fully simulated prototype with projection mapping, so the installation could be experienced before it was built.
Built for the successful pitch of Feast (2021), WangShui's commission for the Hammer Museum's No Humans Involved, Los Angeles (2021 to 2022).
MayaUnitywalkable simulationprojection mapping
Teaching2024 · Woodbury University (now part of University of Redlands)
GAME 203: Sound Design Fundamentals
A university course I created and teach on designing sound and music for games.
Students leave able to build working game audio in a real game engine, not just talk about it.
An analysis-to-CRM pipeline that extracts structured data from long public documents and ranks it into scored opportunity lists.
Converted 50+ unstructured datasets from 20+ districts into scored, ranked opportunities, replacing hours of manual reading with a repeatable pipeline.
PythonJSON schemaScoring rubricCRM
Audio & Music2018 to 2026 · Credits
Game & Film Credits
Music and sound across 17 released game titles and 14 film and TV projects.
31 released credits spanning games, film, and television.
Roles across applied AI, full-stack engineering, hardware, and teaching.
Treobytes, Product Development Specialist San Diego · Aug 2025 to Present
Designed and shipped a production Claude AI agent system with RAG and an automated LLM-as-judge review that cut document turnaround from four days to five hours. Built and maintain TreoTools, the internal React platform adopted org-wide, and designed a buildable Arduino MIDI kit.
Woodbury University (now part of University of Redlands), Adjunct Professor, Game Art & Design Burbank · Aug 2024 to Present
Designed and teach GAME 203: Sound Design Fundamentals, taking undergraduates from DAW fundamentals to adaptive game audio in Unity and Unreal Engine 5 with Wwise and FMOD middleware.
Treobytes, Summer STEM Facilitator & Program Manager San Diego · Jun 2020 to Aug 2025
Built technical curricula from the ground up; taught programming (Python and C++), robotics, and circuitry to 200+ students; led summer STEM operations across five sites and trained and onboarded 30+ facilitators.
Freelance, Full-Stack & Client Engineering Los Angeles / remote · 2022 to Present
Front-end and technical audio work for clients including AltaMed, where I contributed to a shipped English/Spanish healthcare translator tool, and indie game studios, including UE5 adaptive-audio tooling later showcased at GDC San Francisco 2024.
About
About Vi
I'm an engineer, hardware developer, and composer in Little Italy, San Diego. I build things end to end: the code, the circuit board, the CAD, the music, and the curriculum to teach all of it.
At Treobytes I designed and shipped a production Claude AI agent system, solo, that turns live data into compliant, submission-ready documents with RAG and an automated LLM-as-judge review, cutting turnaround from four days to five hours. I built the team's internal React platform from the ground up, wrote analysis pipelines that turn unstructured documents into scored prospects, and designed an Arduino MIDI instrument as a buildable STEM kit.
I've showcased an adaptive game-audio tool at GDC San Francisco, scored 17 games and 14 film and TV projects, earning Silver for Best Composer at the 2022 Tokyo Film Awards, and seen my computational design research featured in Dezeen. I hold a B.S. in Applied Computer Science (Media Arts) from Woodbury University, Magna Cum Laude with a 3.8 GPA and the Best Thesis Award, and I teach GAME 203: Sound Design Fundamentals as an adjunct professor.
I'm happiest at the seam between disciplines: applied AI inside a real product, or audio code talking to a circuit board. I'm currently building CADence, a browser 3D modeler.
Best Composer (Silver), Tokyo Film Awards 2022GDC San Francisco 2024Dezeen 2023Magna Cum Laude · Best Thesis17 games · 14 film/TV
Press & Awards
Recognition
Silver, Best Composer, Tokyo Film Awards 2022
Awarded for the original score of the short film HAIKU. The film also took a Best Original Music nomination at the 2022 New Creators Film Awards and was featured in IndieWrap's 2023 published short-film collection. IMDb
Showcased at GDC San Francisco 2024
A generative game-audio tool, built in Unreal Engine 5 for an indie game studio, was showcased at the Game Developers Conference in San Francisco. It assembles a unique soundtrack per game session from interchangeable musical layers.
Featured in Dezeen, 2023
Computational design work was featured in Dezeen's School Shows. The generative piece computes natural numbers and runs an L-system through multiple iterations to grow fractal forms in the browser. Dezeen · Live demo
Behind Feast at the Hammer Museum, 2021
In December 2020 I modeled WangShui's sculpture concepts and built the gallery-simulation videos used to pitch Feast (2021), commissioned for No Humans Involved at the Hammer Museum, Los Angeles, curated by Erin Christovale. Exhibition
Magna Cum Laude, Woodbury University
B.S. in Applied Computer Science (Media Arts), 2023, with a 3.8 GPA. Recognized with the Best Thesis Award for the graduating class, Academic Excellence Awards in 2022 and 2023, and Dean's List every semester.
Contact
Let's build something
Open to engineering and product roles across California or remote. If you're working at the intersection of software, hardware, and human craft, I'd love to talk.