Research Projects

See below for more information about all of the Summer 2026 projects.

Project 1 - World Agents for Accessibility

Project 2 - Mentra: Adaptive LLM Systems to Support Individuals with Intellectual and Developmental Disabilities

Project 3 - Analyzing The Impact of Mobile Tutoring and Goal Setting on Middle School Math Learning

Project 4 - Strengthening Peer-Run Safety-Net Behavioral Health Services through Community-centered AI Design

Project 5 - Community-Driven Simulation and Robot Learning for Socially Compliant Public Robots

Project 6 - Examining Safety Culture in the Automation of Public Transit

Project 7 - Active Learning in STEM Education

Project 8 - Long-term Impacts of AI Robotics Summer Camp

Project 9 - A Real‑Time Collective Access Agent for Inclusive Board Game Play

Project 10 - Designing Creativity Support Tools for AI Red Teaming

Project 12 - Redrawing the Boundaries of AI Acceptance in Human Conversations

Project 13 - Designing for Worker-Consumer Solidarity in AI-Mediated Labor Markets

Project 14 - Visualizing AI for Treatment Decisions in Healthcare 

Project 15 - Human-Centered Data Science and Visualization 

Project 16 - Comprehensive Personalized Programming Practice Environment

Project 1
World Agents for Accessibility

Mentor: Jeff Bigham, faculty

Description: We are exploring how frontier AI model capabilities can be brought to bear on long-standing accessibility challenges experienced by people with disabilities in the real world. This project will involve incorporating context from the environment (audio, video, etc), processed and acted upon by AI models, and presented in a useful way to solve accessibility problems. For instance, we might explore how visual context can be useful described to people who are blind, or how aural speech context can help a person with cognitive disabilities review what was said during a lecture. We are especially interested in combining these aspects with more agentic aspects, such as systems that proactively describe content of interest or that call appropriate tools to process or transform context to be useful in the moment for the tasks at hand.

Nature of Student Involvement: Students will be involved in prototyping and evaluating systems in the world model space.

Skills we are interested in (not all required):

• Strong technical prototyping skills (able to work with a wide variety of different programming languages, frameworks, including hardware)
• Familiarity with AI models (bonus: experience fine-tuning / training own models)
• Familiarity with human-centered design (bonus: familiarity with accessibility)

_____ 

Project 2 
Mentra: Adaptive LLM Systems to Support Individuals with Intellectual and Developmental Disabilities

Mentor: Hamza El Alaoui, graduate student advised by faculty Jeff Bigham

Description: This project explores how large language models (LLMs) can be adapted to assist intellectual and developmental disabilities (IDD) in everyday communication, writing, and task planning. Building on an existing research prototype, the project will focus on creating a personalized and context-aware conversational interface that dynamically scaffolds users’ cognitive processes. The system will iteratively evolve through participatory design with the target population, ensuring accessibility, transparency, and trust. The student will collaborate with the PI to conduct co-design sessions, implement feedback-driven design changes, and evaluate the system’s impact on usability and autonomy. The goal is to advance our understanding of how generative AI can meaningfully augment human cognition and support independence, leading to a CHI 2027 publication.

Nature of Student Involvement: The student will play an active role in both research and development. They will be assigned readings on LLMs and cognitive accessibility, assist in user studies with individuals with IDD, and write code. They will extend and refine an existing system using ReactJS and backend APIs, implementing features based on user feedback from iterative testing. The student will also contribute to the preparation of study materials and, where appropriate, assist in co-authoring the publication resulting from the project.

Skills we are interested in (not all required):

• Experience with web development (JavaScript, ReactJS or similar frameworks).
• Interest in accessibility, participatory design, human augmentation, or human-centered AI.
• Strong written and interpersonal communication skills for working with diverse user populations.
• Willingness, empathy, and adaptability when collaborating with participants from diverse backgrounds and abilities, including participation in in-person user studies as needed.
• (Preferred) Familiarity with LLM APIs or LLM integration (e.g., OpenAI API, LangChain).
• (Preferred) Prior experience with user studies or HCI research.

_____ 

Project 3
Analyzing The Impact of Mobile Tutoring and Goal Setting on Middle School Math Learning

Mentors: Conrad Borchers, graduate student advised by Vincent Aleven and Ken Koedinger 
Paulo Carvalho, faculty

Description: We are analyzing data from classroom experiments that evaluate the effects of different homework interventions on student learning gains and learning rates within intelligent tutoring systems. These interventions include (1) smartphone-based tutoring with offline capabilities that extends instructional support beyond the classroom, and (2) a goal-setting dashboard that provides students with feedback on their effort and progress, combined with incentives for goal completion.

We are seeking 1-2 students with strong data analysis skills in R or Python (ideally both), and solid understanding of inferential statistics and modeling. In this role, you will contribute to research reports on intervention effects and have the opportunity to pursue independent research questions with mentorship from graduate students and postdoctoral researchers. Potential directions include analyzing step-level engagement patterns, usage of offline tutoring outside school, and trajectories of effort regulation during practice.

Nature of Student Involvement: Students will support data cleaning, statistical modeling, and visualization of learning analytics data from classroom-based tutoring system deployments. They will work closely with graduate student and postdoctoral mentors to analyze the impact of homework interventions on learning outcomes and process-level behaviors. Students will contribute to written research summaries and may pursue independent research questions (e.g., patterns of out-of-class practice or changes in effort regulation over time), with opportunities for co-authorship on publications subject to interest and contribution.

Skills we are interested in (not all required):

• Experience with data analysis in R or Python (both a plus)
• Coursework or background in statistics, data science, or learning analytics
• Ability to work with messy, real-world datasets and apply appropriate statistical models
• Strong communication skills, including writing clear summaries of findings
• Interest in education, intelligent tutoring systems, or learning sciences research

_____ 

Project 4
Strengthening Peer-Run Safety-Net Behavioral Health Services through Community-centered AI Design

Mentor: Hong Shen, faculty

Description: Conducting participatory design activities with community partners to develop and evaluate LLM-powered tools.

Nature of Student Involvement: Students will work with applied AI researchers and community members to design and evaluate LLM-powered tools.

Skills we are interested in (not all required):

• Interested in HCI and design research, including conducting user studies and co-design workshops
• Some experience with LLM is preferred

_____ 

Project 5
Community-Driven Simulation and Robot Learning for Socially Compliant Public Robots

Mentor: Nik Martelaro, faculty

Description: This project investigates how community-driven feedback can improve the social navigation and communication of service robots operating in public spaces. We aim to extend a state-of-the-art open-source robot simulator (SEAN 2.0) to model diverse pedestrians, including wheelchair users and scooter riders, and to collect participatory data from community members. Students will help develop simulation environments, integrate human preference data, and train robot navigation models that balance safety, efficiency, and social compliance. The project connects human-centered design and robot learning by enabling participants to visualize and evaluate real-world human–robot interactions. The resulting tools and datasets will support broader research in socially aware navigation, accessibility, and inclusive robot deployment in public environments.

Nature of Student Involvement: Students will collaborate on developing and testing robot simulation environments, assist in data collection from participatory studies, and train or fine-tune navigation models using reinforcement or imitation learning. They will have the opportunity to contribute to both the human-centered design aspects (such as scenario design and environment modeling) and the technical development side (simulation scripting, model training, and robot testing). Students will also be encouraged to co-author papers and contribute to open-source releases that advance accessible and socially compliant public robotics.

Skills we are interested in (not all required):

• Proficiency in ROS or Unity is required; Experience with both is a strong plus.
• Familiarity with robot learning; Experience training reinforcement learning–based navigation models is a plus.
• Programming skills in Python or C++ and familiarity with simulation frameworks or Gazebo/Isaac Sim are beneficial.
• Experience working with or debugging mobile robots (e.g., AgileX, Fetch, or Unitree) is a plus.
• Interest in human–robot interaction, accessibility, or participatory design is a plus.

_____ 

Project 6
Examining Safety Culture in the Automation of Public Transit

Mentor: Shivani Kapania, graduate student advised by Sarah Fox

Description: This study examines how safety, human factors, and organizational practices intersect in the integration of autonomous vehicle (AV) shuttles and advanced driver assistance systems (ADAS) into public transit. Positioned as tools for enhancing safety, accessibility, and efficiency, these technologies also transform the role of human operators, who must intervene in complex and high-risk scenarios. The project explores how “safety culture” is materialized through the training and everyday practices of both in-vehicle safety operators and remote tele-operators. Safety culture could be understood as the assembly of underlying assumptions, values and attitudes shared by members of an organization, which interact with an organization’s broader contextual setting to result in those readily-visible practices that influence safety.

Nature of Student Involvement: We are looking for a student interested in a qualitative approach to this study. Through interviews with operators, we hope to understand how workers are trained to interpret and respond to safety-critical events. To complement the interviews, we would like to conduct a discourse analysis of video-based training materials to examine how organizational understandings of risk and safety are communicated as part of this “safety culture".

Skills we are interested in (not all required):

• Familiarity with qualitative research methods, including interviewing or thematic analysis
• Interest in the social and organizational dimensions of automation and AI safety
• Good organization skills for planning tasks and communicating project timelines

_____ 

Project 7
Active Learning in STEM Education

Mentor: Paulo Carvalho, faculty

Description: Mastering STEM content requires students to quickly acquire and retain large amounts of complex information, often under time pressure. Technology-enabled learning platforms now allow us to track student performance at scale and deliver practice in adaptive ways rather than relying on fixed schedules. One promising approach is spaced retrieval practice—revisiting questions over time to strengthen memory—but its effectiveness in authentic STEM classrooms remains underexplored. Moreover, most existing systems still use rigid or heuristic spacing rules rather than computational models grounded in how students actually learn. This creates a critical gap: we do not yet know how to optimize learning-by-doing in STEM using data-driven, individualized schedules. What algorithms best adapt spacing to each learner’s history and current state, and how should they be implemented in real classroom technologies?

Nature of Student Involvement: Students will be involved in all steps of experimental research, data analytics, and student modeling.

Skills we are interested in (not all required):

• Quantitative data analyses
• Experience with experimental design and data collection
• Data science/learning analytics/student modeling experience preferred but not required

_____ 

Project 8
Long-term Impacts of AI Robotics Summer Camp

Mentor: Amy Ogan, faculty

Description: AI and robotics are changing the world every day, and the conversation around helping students to have more agency, understanding, and literacy is more important than ever. AI summer camps are a way to think about getting children involved in the conversation and giving them hands-on experience interacting with AI and robots. This project seeks to understand: What are the long-term impacts of these camps, and how do they influence students' feelings about AI and Robotics? Also, how does the camp affect their learning?

Nature of Student Involvement: The student will be actively assisting with the facilitation of the AI robotics summer camp by helping to prepare material and study design working directly with middle school students and observing their learning in real-time. They will also be a key part of helping to process the data collected with an opportunity for publication.

Skills we are interested in (not all required):

• Qualitative data analysis
• Study design
• Interest or experience with teaching children about technology

_____ 

Project 9
A Real‑Time Collective Access Agent for Inclusive Board Game Play

Mentor: Joon Jang, graduate student advised by Andrew Begel and Patrick Carrington    

Description: This project designs and evaluates a table‑centered agent that helps small groups uphold co‑authored communication practices (e.g., periodic recaps, one‑at‑a‑time speaking) when playing board games. The agent listens via tabletop microphones, detects audio patterns and issues prompts that participants pre‑authorize. We combine participatory co‑design and controlled lab sessions to study which designed interventions improve inclusion and enculturation, and whether effects persist without the system. Students can get involved with prototype audio, UI, haptic features, support study operations, and analyze mixed‑methods data. Their work will blend HCI, accessibility, conversational agents research, and students can get experience with developing complex ASR applications with rule‑based logic.

Nature of Student Involvement: Prototype additional features in Python/Arduino for the tabletop system. Build and test automated and semi‑automated study tools; maintain a control panel for pause/override for the system during study sessions. Prepare and assist in‑person sessions; set up systems, monitor sessions, and collect short surveys. Analyze data qualitative and quantitative data resulting from participant sessions. Contribute to weekly research meetings, literature reviews, and an end‑of‑summer poster. Possibility of participation in writing a paper draft with authorship opportunities.

Skills we are interested in (not all required):

• At least one of: (a) audio/ASR or DSP experience (PyAudio/Whisper), (b) embedded prototyping (Arduino/BLE), or (c) qualitative HCI methods (interviews/coding).
• Solid Python skills; willingness to learn light web/Node and device networking.
• Interest in disability/accessibility.
• Preferred: experience with real-time systems, basic hardware wiring, or prior user study assistance.
• Bonus: familiarity with conversation analysis concepts (turn‑taking, repair); enthusiasm for board games!

_____ 

Project 10
Designing Creativity Support Tools for AI Red Teaming

Mentor: Alice Qian Zhang, graduate student advised by Hong Shen and Laura Dabbish

Description: AI red teaming, the practice of stress-testing AI systems by finding creative ways to break them, is crucial for AI safety. Red teamers need to think outside the box to uncover vulnerabilities that developers missed, but this creative work can be challenging and mentally taxing. How can we better support red teamers' creativity and help them discover novel risks in systems more effectively? This project uses co-design methods to develop creativity support tools specifically for AI red teaming. We'll run collaborative workshops with experienced red teamers to understand their creative processes, identify pain points, and iteratively design tools that enhance their ability to find system vulnerabilities and safety issues.

Nature of Student Involvement: What we expect from a student: Help design and facilitate co-design workshops with AI red teaming professionals, analyze workshop recordings and notes to identify design insights, prototype creativity support tools based on participant feedback, conduct user research to understand red teamers' workflows and needs, and contribute to developing novel interfaces that support creative security testing.

Skills we are interested in:

• Required:
  • Strong academic performance as indicated through GPA or other academic work
  • Strong communication and interpersonal skills for workshop facilitation
  • Attention to detail
  • Experience with qualitative research methods
• Preferred:
  • Experience with prototyping tools (Figma, Adobe Creative Suite, or similar)
  • Background in HCI, design, psychology, or related fields
  • Prior experience with user research methods or workshop facilitation
  • Basic understanding of AI systems and AI red teaming

_____ 

Project 11
Community-Driven Simulation and Robot Learning for Socially Compliant Public Robots

Mentor: Nik Martelaro, faculty

This listing (Project 11) is a duplicate of Project 5 above. If you are interested in working on this project with Nik, please select Project 5 in your rankings.

_____ 

Project 12
Redrawing the Boundaries of AI Acceptance in Human Conversations

Mentor: Hirokazu Shirado, faculty

Description: This project investigates how people decide where and when it is acceptable to use AI assistance in interpersonal conversations, ranging from purely transactional contexts (e.g., scheduling) to deeply social ones (e.g., emotional discussions with partners). Using an online vignette experiment, participants will evaluate the appropriateness of AI involvement under varying levels of intimacy, capability, and governance. Beyond measuring current attitudes, the study explores how potential “what-if” features—such as AI systems that could form intimate relationships or hold an economic stake—might shift these boundaries of acceptance. The findings will deepen our understanding of human rationales for AI acceptance and inform the design of AI assistance systems that align with evolving social and moral expectations.

Nature of Student Involvement: Students will be involved in all stages of the research process. They will help design and pilot online vignette surveys, create and refine realistic conversational scenarios, and conduct both quantitative and qualitative analyses of participants’ responses. Students will learn to use tools for experimental deployment and R or Python for data analysis and visualization. They will also participate in weekly lab meetings, contribute to interpreting results, and may co-author conference submissions depending on their level of contribution. This project provides hands-on experience in computational social science, AI ethics, and human–computer interaction research.

Skills we are interested in (not all required):

• Required: Interest in human–AI interaction, social computing, or moral psychology.
• Familiarity with basic survey design or social science research methods.
• Experience with online survey tools or willingness to learn quickly.
• Basic data analysis skills in R or Python.

_____ 

Project 13
Designing for Worker-Consumer Solidarity in AI-Mediated Labor Markets

Mentor: Jane Hsieh, S3D PhD Graduate

Description: Since their emergence a decade and a half ago, gig platforms have transformed relations between consumers and workers across a wide range of services. Today, digital and AI-mediated technologies mediate more and more of our everyday service interactions, including food delivery and ridehailing. However, the underlying mechanisms of production and consumption remain the same: workers continue to produce services and data that platforms exploit to consolidate increasingly globalized power, while consumers continue to (de)value these services and products through monetary means.

Previous HCI research has revealed oppressive platform practices that disempower workers through data extraction, low pay and exploitative conditions – but this worker-centered lens has largely overlooked how platforms also reduce consumer agency and control, limiting opportunities for consumers to provide and express social support and advocacy for workers. Gig platforms further blur these roles by legally classifying workers as consumers who use their matching algorithms and networks. While platform designers and workers themselves have limited leverage to reshape worker-platform interactions in ways that empower workers, consumers carry potential for (and often ideological motives) influencing platform reputations and practices. In this comparative study, we explore how interaction probes might scalably and sustainably motivate consumer boycotting behaviors across two domains: food delivery, a three-sided market consisting of restaurants, workers and consumers; and ridehailing, a two-sided market with only workers and consumers.

Nature of Student Involvement: The student will participate in at least the design and development phases of an interaction probe, which will require in-depth understanding of user stakeholders involved. The student may also optionally engage in the evaluation and reporting stages of the study.

Skills we are interested in (not all required):

• The student should have interest in computing research and labor studies
• Experience with lit reviews and qualitative methods is preferable
• Interest in service or interaction design are desirable

_____ 

Project 14
Visualizing AI for Treatment Decisions in Healthcare

Mentor: Adam Perer, faculty

Description: We are working on building interactive visualization tools to help physicians and other healthcare providers diagnose and treat diseases more effectively using artificial intelligence (AI) algorithms. These tools leverage historical patient trajectories to identify and recommend optimal treatment plans to clinicians in real-time. However, providing these recommendations to clinicians is no simple task - clinicians often dismiss them as untrustworthy, or pick and choose aspects of the recommendation to follow. We are looking for a motivated student to work with us and our clinical collaborators to understand how best to deliver AI-generated recommendations.

Nature of Student Involvement: Students will interact with healthcare experts and design new interactive, visual interfaces.

Skills we are interested in (not all required): 

• Interest in healthcare, data visualization, and web-based programming.

_____ 

Project 15 
Human-Centered Data Science and Visualization

Mentor: Adam Perer, faculty

Description: The CMU Data Interaction Group (DIG) has a mission to empower everyone to analyze and communicate data with interactive systems. Our group conducts research in computer science at the intersection of human-computer interaction, machine learning, data science, programming languages, and data management. This summer, we plan to build new tools for data scientists to help them better understand their data, which will hopefully result in better downstream machine-learning models derived from the data.

Nature of Student Involvement: Research, coding, and user studies.

Skills we are interested in (not all required):

• Programming experience
• Interest in data science and machine learning
• Web development skills

_____ 

Project 16 
Comprehensive Personalized Programming Practice Environment

Note: If you are interested in this project, please select project 7 on the ETAP application

Mentor: Paulo Carvalho, faculty

Description: Computer science education is increasingly critical for preparing well-trained professionals for the national economy and building a competitive workforce of the future. This project capitalizes on the power of generative AI and progress in learning science research to provide personalized learning experiences for computer science students. The system will recommend the most suitable learning activities for each student according to their current knowledge level and offer personalized feedback to support their progress. A sequence of experiments will lead to a better understanding of the kinds of practice opportunities (i.e., worked examples vs problems) and types of feedback messages that are most effective to each student depending on their prior knowledge and interest.

Nature of Student Involvement: Students will be involved in all steps of experimental research, data analytics, student modeling, and LLM-based material development.

Skills we are interested in (not all required):

• Quantitative data analyses
• Experience with experimental design and data collection
• Data science/learning analytics/student modeling experience preferred but not required
• Experience with LLMs preferred but not required