Projects
A list of active projects we're working on.
Co-pilot Grading Tool 
The Copilot Grading Tool (CGT) is a web plugin designed to assist teachers and teaching assistants in grading student work more efficiently. Powered by advanced Large Language Models (LLMs), the tool analyzes student submissions and delivers detailed feedback aligned with the instructorâs grading rubric. CGT is built to seamlessly integrate with widely-used learning management systems (LMS) such as Canvas, ensuring easy incorporation into existing workflows.
The priority of CGT is ensuring privacy, ethical, and responsible use of AI. The tool anonymizes all student submissions before processing and does not store any student data on remote servers.
Check out an early prototype of CGT in action (gif)
Prompt Engineering Education
Prompt Engineering is the process of designing and refining the input prompts given to LLMs to achieve specific outcomes. We are exploring the use of Prompt Engineering to improve the performance of Large Language Models (LLMs) in educational settings. In the context of education, we are interested in using Prompt Engineering to improve the quality of generated responses to student queries, feedback, and assessments.
The priority of this project to develop a set of best practices for pre-processing and structuring prompts to maximize the accuracy and relevance of AI-generated materials in educational contexts. This project is at the heart of all projects at Learnification Labs.
Early experiments have led us to develop BugsyPrompt.
Automated PrairieLearn Question Generation
PrairieLearn (PL) is an online assessment and learning system that empowers instructors to create robust educational resources for students. In the process of developing a new Web Development PL-based course, weâre exploring the integration of LLMs to develop a new pipeline for generating personalized feedback and hints for students.
The priority of this project is to improve accuracy and relevance of AI-generated materials in our PL-based course pipeline.
Edu-val: Peer Review and Evaluation
Canvasâs built-in peer review feature has many limitations. Specifically, it lacks the ability to assign peer reviews for group-based assessments, which is a common practice in many educational settings. Additionally, the feature does not allow for any peer evaluation or grading, which is an essential aspect of group-based learning.
Educational Evaluation (Edu-val) is a web-based platform to solve these shortcomings. This tool is designed to streamline the peer review and evaluation process for group-based assignments. The tool is built to resemble canvas and allow students to peer review each otherâs work, provide feedback, and evaluate their peersâ submissions.
The goal of this project to improve the peer review and evaluation process for group-based assignments in an educational settings.
AI-driven Computational Thinking Exercises
Bebras challenges are an internationally recognized set of Computational Thinking (CT) exercises designed to engage students in problem-solving and algorithmic thinking. We are exploring the use of Large Language Models (LLMs) to generate new Bebras-like questions that are accurate, contextualized, and appropriate for students at different levels of education.
A project goal is to develop a set of guidelines to assess LLM-based CT questions and develop a pipeline for generating new questions that align closely with Bebras challenge standards.
Gamified Geeks: Practice Coding Platform
Gamified Geeks (GG) is an interactive learning platform that combines gamification pedagogy with AI-driven question generation to allow students to practice coding exercises in a fun and engaging way. The platform offers a variety of coding challenges, ranging from beginner to advanced levels, and provides instant feedback to help students improve their coding skills.
The project focus is on fine-turning the LLM-based question generation system with a gamified interface to promote active learning and engagement among students.
Version-wise: Learning Git Made Fun
LearningGitBranching is a web-based platform that provides an interactive and engaging way to learn Git. The platform offers a variety of commands and challenges to help students understand the fundamentals of Git, such as branching, merging, and rebasing.
One of the limitations of the platform is the lack of personalized learning features. With Version-wise, we are exploring the integration of Large Language Models (LLMs) to provide personalized feedback and hints to students based on their progress and performance. The goal is to improve the learning experience and effectiveness of the platform by tailoring the content to individual student needs.
Technology Stack
Most, if not all, of our projects are built using MERN stack (MySQL, Express.js, React, Node.js).