During my internship and thesis, I contributed to the ongoing research project “Participatory Observation and Bias Testing of Large Language Models”. The project aims to enable users to engage with synthetic personas for reflective critique of emerging technologies in relation to design inclusivity, fostering awareness of bias and representation in artificial intelligence systems.
My work focused on designing and developing a research portfolio website and leading the UX design of an interactive persona-based Bias Tool, positioning design as both a communication medium and a research instrument.
Duration
6 month
Context
This project was conducted within the Department of Information Engineering and Computer Science at the University of Trento, in collaboration with a PhD researcher and under academic supervision.
Role
I worked across design + implementation, with a strong focus on usability and interaction flow.
Responsibilities included
UX/UI Design & Interaction Design
User-centered + participatory design activities
Prototyping (low → high fidelity)
Usability testing + analysis (qualitative + lightweight quantitative)
Web design + implementation
LLM integration proof-of-concept (front-end ↔ model endpoint)
Part 1: Research Portfolio Website
The first phase of the internship focused on designing and implementing the researcher’s personal website as a digital platform to present publications, research themes, and project outcomes. The website was built using WordPress and Hostinger Cloud, following specific technical and hosting requirements.
I focused on:
information architecture (making complex research easier to navigate)
visual hierarchy & readability
a structure that connects the research story to the tool
The website functions as the entry point and framing layer for the Bias Tool (tool access restricted for research/publication reasons).
Part 2: Persona-based bias tool
QUID/AI, the Bias Tool was designed to enable users to interact with synthetic personas in order to reflect on representation, bias, and inclusivity in emerging technologies. The tool is grounded in the researcher’s empirical PhD work and is intended to support critical engagement with carefully selected personas.
Due to its conceptual and methodological complexity, the tool was designed using a hybrid approach combining participatory design (PD) and user-centered design (UCD).
Onboarding Screen
persona detailed information
Chatting with a fictional persona
Chatting with a fictional persona
persona detailed information
We followed a theory-driven, human-centered design process, relying on secondary research to ground our design decisions in existing evidence and best practices.
Discover
I began by working closely with the researcher to understand the theoretical and empirical foundations of the project, alongside its ethical and methodological constraints. This phase also included secondary research on persona-based evaluation, chat-based interfaces, and existing LLM interaction patterns.
Define
Translated research requirements into concrete UX goals, such as:
making persona selection understandable and purposeful
supporting quick scanning + deeper exploration of persona profiles
making multi-persona interaction clear (switching/opening chats, continuity, deletion expectations)
Develop
Sketched flows and iterated through wireframes based on the researcher's feedback and workshop sessions.
Designed a high-fidelity prototype (Figma) and refined micro-interactions
Iterated specifically on persona navigation, profile discoverability, and onboarding clarity
Evaluate and Iterate
Ran usability testing focused on UI clarity, interaction flow, and navigation (not conversation quality)
Used task-based testing + short follow-up questions + SUS
Alongside the design prototype, I supported a functional proof-of-concept of the tool as a web app. The system was implemented using HTML and JavaScript and was integrated with an LLM endpoint and deployed across hosting/proxy infrastructure.
(Access remains restricted to support ongoing academic work.)
Turned research requirements into a usable product experience through iterative prototyping and user testing.
Used evidence from usability sessions to refine onboarding, persona discovery, and multi-chat navigation.
Practiced HCI-centered design under real constraints: ethics, expectations, and research validity.