An adaptive, multilingual communication platform for non-verbal and low-literacy users.

Developed with the ACE Lab (UC Berkeley), this project combines animated avatars, context-aware phrase prediction, and multilingual text-to-speech to support expressive, accessible communication. The system learns user context to generate relevant suggestions and visual cues, reducing cognitive load and enhancing independence in daily interactions.

Built with React, Flask, and OpenAI APIs, the tool emphasizes accessibility, cultural inclusivity, and emotional resonance in AAC design. Currently in prototype development.

(Link Coming Soon)

Automated reminders for students, powered by a single source of truth.

Developed in Dr. Dan Garcia’s ACE Lab (UC Berkeley EECS), AutoRemind is a microservice that centralizes coursework communication. Integrated with GradeView, it automatically delivers email and SMS reminders for assignments, announcements, and updates. Built with Next.js, Supabase, Twilio, and SendGrid, the system ensures consistent, timely messaging to help students stay organized and engaged.

Preview Website (Development in Progress)

Gave computer vision systems object permanence, the ability to recognize hidden objects, and applied it to autonomous driving

Collaborated with a cross-UC team at LLNL’s 2025 Data Science Challenge to develop U-Net and ConvLSTM models in PyTorch for detecting occluded objects. Improved object permanence with RGB/video tracking, achieving strong F1 (0.84) and IoU (0.79) scores on synthetic and real driving data. Presented results via research poster and technical notebook with applications in robotics and autonomous vehicles.

Explore Our Poster

Explore Our Code

An AI-powered project management chatbot to generate context-aware documentation. 

Built an AI-powered project management chatbot that generates context-aware documentation, leading a five-member team, securing a UCR research mini-grant, and earning recognition through the Pillars of Excellence Award, $3,800+ in scholarships, and a semi-finalist finish in the 2025 Mays Business School AI Pitch Competition. Developed with Node.js, HTML, CSS, and the OpenAI API; the functional proof-of-concept site (inactive since June 2025) remains available for preview.

Preview Website

Design guidelines for medical conversational agents that support elderly individuals.

In collaboration with Dr. Sanjoy Moulik, this research project focused on designing a chatbot to assist elderly patients in managing chronic disease symptoms. Key activities included collecting and annotating research articles and developing chatbot prototypes using various interfaces. The data collection and prototype phases have been successfully completed, and the project is now awaiting IRB approval for human subject testing to advance further development.

Publication in Progress:

Moulik, S., & Santana, L. (2025). Design principles: Conversational agents for chronic disease management.

A clinical research tool to help researchers find patient groups and plan studies.

Developed with the Director of Scholarly Activities in the School of Medicine, this tool leverages real-time patient data to streamline clinical research. It enables quick cohort identification, feasibility assessments, and study design using advanced analytics. The project, Novak, D., & Santana, L. (2025, February). Emerging Technologies for Teaching and Learning: Digital Demonstrations, will be presented at the Association of American Medical Colleges.

ArcGIS web app visualizing U.S. election results and voting trends.

This interactive web app, built using ArcGIS, visualizes and compares election data across multiple layers. It includes a map comparing the 2020 Presidential and 2022 Midterm election results, providing insights into voting trends over time. Additionally, the app features a detailed map of the 2020 Presidential election results in Illinois, analyzed alongside the state's 118th congressional districts. This project highlights geographic patterns in voter behavior and serves as a valuable tool for exploring electoral shifts and district-level outcomes.

ArcGIS web app tracking live snowplow movements in real time.

This project utilizes real-time mapping to track live snowplow movements, with truck locations represented by dynamic icons and their paths marked with colored lines. The interactive visualization provides clear insights into snowplow operations, enabling users to monitor coverage, optimize routes, and improve response efficiency during snow events.

ArcGIS web app highlighting communities with low educational attainment.

This project features an interactive map designed to identify areas most in need of afterschool programs by visualizing educational attainment and equity indicators. The map highlights regions where the percentage of the population aged 25 and over with less than a high school education exceeds national averages. Darker purple shades indicate higher percentages, with the scale ranging from below 1% to over 21%. This tool provides valuable insights to help target resources and improve educational equity in underserved communities.

Check out some of my featured projects that reflect my work in AI, data science, and educational technology:

• U.S. Air Quality Trends Analysis (2000–2023)

• Reducing Bias in ML for Depression Prediction

• NLP Curse Word Detection

• AI Education Impact Study

• Data Science Ethics

🔗 Explore my full GitHub Profile