Back to Resume

Homework Hotline

Bilingual Homework Hotline: AI-Powered Tutoring Support

4 Years Running AI-Powered 3 Institutions

In 2020, as schools shifted online and families navigated unprecedented uncertainty, a group from the University of North Texas, Texas Woman's University, and Denton ISD began meeting weekly to ask a simple but urgent question: How can we help bilingual families and students succeed?

From those conversations, the Bilingual Homework Hotline was born. Each week, pre-service teachers from both universities provide one-on-one and small-group tutoring for K-12 emergent bilingual students via Zoom, applying bilingual and ESL instructional strategies in a real-world setting.

Now in its fourth year, the program serves over a hundred students each semester while preparing the next generation of bilingual and ESL educators.

Impact Numbers

4 Years of continuous operation
3 Partnering institutions (Denton ISD, UNT, TWU)
100+ Students served each semester
2 AI chatbots built (tutoring + onboarding)

My Role: Designed and built the project website, the AI Homework Helper chatbot (RAG-powered, bilingual), and the Onboarding Assistant. I also maintained the centralized resource hub and integrated AI tools into the tutoring workflow.

What I Built

Two AI chatbots serving different needs within the same program

HH Homework Helper (Text Chatbot)

RAG-powered chatbot that helps tutors scaffold explanations instead of giving direct answers. Uses bilingual education best practices, cognate awareness, and culturally responsive strategies. Supports image generation for visual learners.

GPT-4o RAG SSE Streaming DALL-E 3 3-Layer Safety

Onboarding Assistant

Conversational AI agent that answers new tutor questions about the program, teaching strategies, Zoom tips, and expectations. Supports bilingual interaction in English and Spanish, with program knowledge built directly into the system prompt.

GPT-4o Bilingual SSE Streaming Program Knowledge

3-Layer Content Safety

Regex pattern filtering, OpenAI Moderation API, and system-prompt guardrails ensure safe, on-topic interactions. Designed for a K-12 educational environment where content safety is non-negotiable.

Regex Moderation API Prompt Guardrails

Static Site + API Architecture

Migrated from WordPress to a lightweight static HTML site with Node.js/Express API backend. Custom domain, SSL, and nginx reverse proxy with redirect mapping for all legacy WordPress URLs.

Node.js nginx Let's Encrypt PM2

The Story

From pandemic response to a lasting ecosystem of learning and mentorship

While serving as a Visiting Assistant Professor of Bilingual Education and ESL instruction at Texas Woman's University, I was invited by my colleague Dr. Jorge Figueroa to collaborate and integrate my students into the Bilingual Homework Hotline initiative. A clear pattern quickly emerged: tutors found it challenging to respond to specific questions, particularly in math and science, that required both accurate content knowledge and the ability to explain it at different language proficiency levels.

To bridge this gap, I built a centralized website that served as both a communication hub for tutors and a growing repository of instructional materials. When the first large language models emerged in late 2022, I saw enormous potential in integrating them into this educational context.

I began working on two connected ideas: a broader intelligent tutoring framework I would later name Veronica, and a practical application through the Homework Helper chatbot. The Homework Helper became the first working prototype of Veronica's vision, designed to guide tutors through reasoning steps, offer scaffolded explanations in English and Spanish, and suggest strategies aligned with bilingual and ESL pedagogy.

What began as a simple text-based chatbot has evolved into a multimodal learning companion capable of generating images and diagrams to support comprehension. With every new generation of large language models, the system grows more sophisticated, offering richer ways for tutors and students to communicate, visualize, and make sense of ideas together.

Try the Homework Helper

This is the live chatbot tutors use during sessions. It scaffolds explanations instead of giving direct answers, suggests culturally responsive strategies, and supports bilingual interaction. Try it:

Try: "How do I explain fractions to a 3rd grader who speaks Spanish at home?" Try: "A store has 24 apples. If 3 friends share them equally, how many does each get?" Try: "What are good strategies for teaching science to emergent bilingual students?"

Key Design Decisions

01

Never Give Direct Answers

The chatbot helps the tutor teach, not the student cheat. Every response scaffolds understanding through guiding questions and step-by-step strategies.

02

RAG-Powered Accuracy

The Homework Helper uses Retrieval-Augmented Generation with a curated knowledge base of bilingual and ESL best practices, ensuring responses are grounded in proven pedagogical strategies.

03

Bilingualism as Strength

The system treats bilingualism as an asset, not a deficit. It suggests cognates, culturally relevant examples, and validates home languages throughout every interaction.

04

Education First, Technology Second

AI enhances human teaching, it does not replace it. The technology handles routine tasks so tutors can focus on listening, motivating, and building relationships.

¡Solo es posible juntos!

It is only possible together. The Bilingual Homework Hotline continues to grow through the collective effort of administrators, teachers, university faculty, and students.