Portfolio

Mollie E. Muñoz

AI Interaction Designer  ·  Information Designer  ·  Software Engineer

Designing the space between human and AI.

AI Interaction Design

2022 — Present

My question underneath every project: who is on the other side of this interaction, and what do they need to be more fully themselves? These projects span modalities, constraints, and years — but they're all attempts at the same thing.

September 2024 Hackathon · Spring 2026

Reading Level Web Translator

Screen shot of a Wikipedia page with the paragraphs being translated to a different reading level

This project is for children and adults, second language learners, and anyone who shouldn't need to go to a different application to access the same information designed for the default audience.

The goal — a browser extension that renders an existing webpage at any reading level, for free, starting with Wikipedia.

The new extension has caching, progressive processing, and runs Llama 3.2 3B locally.

Original Demo → New Demo → New extension →
Meet people where they are. Then help them grow.
Summer 2025

Dual-Modality Document Generator

UI for an app to generate documents using prompts and manual text

Investigated how a document is decomposed, such that it can be built back up again using multiple sub-component prompts and text.

The prototype was then extended to explore a chat-driven application. Agent tool calls replaced traditional user interactions such as clicks to accomplish desired outcomes.

Gradio was used as a prototyping trial, but wasn't the right fit. The model had limited training on the framework and deployment faced DOM differences.

View GitHub → View GitHub with Chat →
My favorite learning was the user feedback — folks had a hard time understanding what to do with the app. I think a better design would have let the user discover the decomposition, rather than introducing them first to unintuitive modules.
September 2022 Hackathon

Spoken Narrative to Structured Documentation

Screen shot of the Spoken Narrative to Structured Documentation app interface

A hackathon project built around a problem that shouldn't exist: engineers who know their systems deeply but have to become professional writers to document them.

Instead, someone speaks naturally — stream of consciousness, jargon, asides and all — and the system transforms that into a structured, formatted guide others can use.

I used Azure Cognitive Services, C#, and GPT-3 with Completions.

View Original Demo →
Patent pending: Enhanced Generation of Formatted and Organized Guides from Unstructured Spoken Narrative Using Large Language Models.
IN PROGRESS: Spring 2026 — Present

AI-Adaptive UI — Personalization by Design

A split tree spanning two different scenes.
In progress
Details upon disclosure

Image: Midjourney

A new AI product concept built around the question: What should an app be like for this person, not for the market persona?

Most software is designed with a specific group of people in mind, but even within that group, needs vary.

Instead of customization options permanently affixed, this project takes a new approach — derive what a user needs from their data and build the experience around that, both persistently and on demand.

AI can generate UI on the fly — we're already seeing that. But people who return to an environment need consistency, not novelty.
IN PROGRESS: Winter 2026 — Present

Reimagining the Human–AI Interaction Contract

Two artists collaborating on a large-scale artwork.
In progress
Details upon disclosure

Image: Midjourney

Artifact creation in the AI era is perhaps at its highest dispense rate. While autonomous output is exciting, have we lost sight of the human process?

Considering how I collaborate with humans in real time and space, I'd like to have a similar experience alongside AI.

This shifts the design problem from output quality to interaction quality and raises harder technical questions about the system architecture.

Patent application in progress — covers a technical design to support this experience.

Thumbnails

2024 — 2025  ·  POCs & quick sketches

Proof-of-concepts that didn't become full projects — the work shifted, someone else picked up the idea, or it solved what it needed to solve and stayed in the margin.

Summer 2025

Parallel Prototyping

I created a script that used Claude Code to generate divergent directions on a problem, then spin up new instances in separate terminals. Each one ran as a subprocess and created whatever artifact it chose to. The challenge was display — the form an artifact would take was unknown.
View on Github →
An AI rendered image of multiple terminals open simultaneously, in a sketched hand drawn style

Image: Midjourney

Fall 2025

Canvas Workspace

I wanted a space I could expand across monitors that tracked all my sessions and work, with notes to self and visual forking. A way to hold the shape of a divergent exploration, not just its outputs. This quick sketch helped me explain the types of features and experience I was looking for.
Click Image to Expand →
I screen shot of a mocked up experience to run multiple AI chat sessions in a canvas with added notes, directory info, and to do list for work
Spring 2024

Spoken Interruption

Humans interrupt each other — we should be able to interrupt an LLM. I wired together Azure Cognitive Speech Services, a chat loop, and GPT-4. When sound was detected, response streaming text output stopped and I appended to the chat history an interrupt-tag for context.
View Demo →
An AI rendered image of two people speaking, represented by chat bubbles, in a sketched hand drawn style

Image: Midjourney

Information Design

2022 — 2023

Designing how knowledge moves from someone who has it to someone who needs it. Different medium, same motivation.

Semantic Kernel video series clip of a person presenting in front of a screen
Fall 2023

Semantic Kernel Series

Independently researched, scripted, and presented a learning arc for technology whose terminology was being defined. Published on the Microsoft Developer Channel.

View series →
An AI rendered image of a child giving a high-five to a robot
Fall 2022 · Published Summer 2023

AI Development Tutorials

Two tutorials for the Maker community on GitHub Copilot. Crafted the information architecture, visual layout, and technical content. Pre-ChatGPT, when AI dev tooling was still nascent.

Article 1 → Article 2 →
A small electronics board sitting on top of multi-colored embroidery thread.
Fall 2022

Azure RTOS for Arduino

A tutorial introducing threading and other real-time operating system concepts to the Maker community. Designed and built both the code and tutorial for Arduino developers.

View on Hackster.io →

Patents

2023 — Present

Work at the intersection of AI systems and human interaction.

  • Granted Real-Time System for Spoken Natural Stylistic Conversations with Large Language Models
    View Patent →
  • Pending Enhanced Generation of Formatted and Organized Guides from Unstructured Spoken Narrative Using Large Language Models
    View Patent Application →
  • Pending Cyclic Behavior Detection in Generative Agents
    View Patent Application →
  • In progress Additional application related to the Human-AI Interaction Contract work.

    Details upon disclosure

Early Work

2005 — 2020

From where I started as a younger adult.

Charcoal drawn seated crossed legs on a light background
2005 – 2008

Charcoal & Film

I worked in charcoal because of the connected movement related to the surface. Film was an exploration of what I could envision and see. Both represented my mark and my presence, no matter how ephemeral.

View gallery →
Newspaper clipping of the Urban Legends of the Fall series, showing a tunnel underground
October 2005

Urban Legends of the Fall

A three-part investigative series with original ideation, research, and writing. Awarded Second Place in the Oregon Newspaper Publishers Association collegiate competition. Published under Mollie Holmes.

Article 1 → Article 2 → Article 3 →
A photograph of a Macy's Juniors department with a brightly painted mural
2009 – 2011

Visual Merchandising

Designed user experiences in retail. Painted murals, transformed departments, and staged experiences at Macy's. Led store to rank 1st in the NW region for percentage-over-goal for in-store event ticket sales.

View gallery →
  • Knowledge Transfer — HP PageWide T100 2015 – 2020 Consolidated and produced documentation across mechanical, firmware, and hardware disciplines for an international manufacturing partner.
  • Mathematics Instruction & Tutoring 2010 – 2013 Adult Learning & Tutoring Certification. Taught college mathematics and customized approaches for non-traditional adult students.
  • The Contemporary Artist's Working Relationships:
    Understanding Restriction and Authority     2008 An undergraduate honors thesis reviewing works of Daniel Buren, Fred Wilson, Andrea Fraser, and Richard Serra.
    View in the OSU Archive →
  • Oregon State University 2008 · 2015 B.S. Mathematics, B.A. Art, B.S. Computer Science

Artist Statement

Thoughts on AI and the human mark

My son is three years old.

When he finishes something — a drawing, a block tower, something he's never done before — he looks up and says I did that. Something that didn't exist before exists now, and no one will ever quite do it the way he did. It took effort, persistence, patience — it took work. And it feels good.

I watch him and I think, we are born with this. But something in us as we age loses touch with it. We live in a world of consumption, of convenience, of outputs delivered instantly without effort. And some people, faced with the choice, decide they would rather have a generic representation of themselves than put in the work of making something their own.

I understand the temptation. I don't want it for my son.

Everything we interact with has the potential to shape how we see the world around us and how we see ourselves. That's why designing for this new interaction space isn't just important, it's one of the most human problems we have ever faced.

We are not interacting with software in its traditional sense. We are not querying a database or navigating a menu. We are in the presence of something that responds, that adapts, that can feel — to us — like another mind. That changes everything. Because how AI interacts with us, or fails to, will shape whether or not we believe our process, our presence, our mark, has value.

I did that. — That's what I'm designing for.

Mollie E. Muñoz