LIZETH ANAYA-OJEDA
LIZETH ANAYA-OJEDA
Ripley, the Casemaker
Designing responsible, inclusive AI
Rales Fellow at Carnegie Mellon University
First-generation, dual master’s student
Raised in the Southwest
Featured Projects
Ripley: AI-Powered Advocacy Platform for WIC Association of New York State
HCI Research, Responsible AI, Agentic Systems
WIC advocates walk into legislative meetings armed with spreadsheets and gut instinct. I helped design a system that changes that — an AI-powered platform that synthesizes community testimony, legislator profiles, and policy data into tailored talking points, rehearses your pitch, and generates slide decks in real time. This is what advocacy infrastructure looks like when you build it from the ground up.
RAG: Retrieval-Augmented Generation for Algorithmic Bias Research
RAG Pipeline, Hybrid Search, Evaluation Design
Research in responsible AI has a credibility problem: the models that could help you navigate the literature will confidently fabricate a citation if you let them. I built a system that wouldn't. A full RAG pipeline over 26 sources on algorithmic bias in public services, with hybrid BM25 and semantic retrieval, 100% citation correctness, and a four-page Streamlit research portal that went from 60% to 100% answer rate across three phases.
Synthetic Data Governance in the Age of LLMs
AI Governance, NIST AI RMF, Empirical Evaluation
FinePhrase was being marketed as privacy-preserving. I wanted to know if that was true. Spoiler: it is not. A two-phase study combining a literature review of 14 papers with an empirical evaluation of FinePhrase's rephrasing pipeline found that synthetic data does not remove PII, it redistributes it in ways that evade standard detection while remaining reconstructable by models trained on it. The governance frameworks meant to catch this have not caught up.
Let’s Connect
I’d love to hear from you! Whether it’s about potential collaborations, questions about my work, or just to say hi.