AI Forge Hero.

AI Forge

Purdue's AI Forge is both a principle and an initiative aimed at providing Purdue's undergraduate and master's students with advanced skills in state-of-the-art generative AI techniques, taught by internationally recognized AI experts. The key principle of AI Forge is giving students agency, knowledge, and resources to allow them to forge their own path in AI. Rather than deciding how students learn, and which projects they need to undertake, as in a typical AI and machine learning course.

AI Forge Modules

A set of short 4-week modules that students can take in nearly any order, with as minimal prerequisites as possible.

  • These short 4-week modules provide students with hands-on knowledge on state-of-the-art foundation models for AI. These modules can be combined into a single course or be taught separately and are initially targeted towards CS/DS/AI undergraduates and first year MSc and PhD students. Later these modules will be opened to the wider Purdue student population and industry partners. These modules emphasize fundamental and practical aspects of generative AI models.
  • Modules are supervised by a research scientist or instruction specialist, along with graduate TAs at a rate of 10 students per GTA, so that students receive close feedback on their assigned projects. GTAs meet twice a week with students and are responsible for ensuring students are making progress in their projects. The modules also have a set of undergraduate TAs that have demonstrated AI experience and have good leadership skills.
  • Modules are designed as hands-on learning, pre-recorded by lead AI faculty.

Modules

Basics of Large Learning Models

Module 1
(description)

In-context learning and RAG

Module 2
(Description)

Multi-modal generative AI models

Module 3
(Description)

Training and inference scalability

Module 4
(Description)

Implicit preference optimization

Module 5
(Description)

Foundation models beyond sequences

Module 6
(Description)

AI Forge Projects

A set of real-world project proposals that students can choose from. These projects can be divided into:

  1. Wild projects: AI applications proposed by any Purdue faculty and staff that are not supervised by them. These projects are supervised by AI Forge's assistant instructors.
  2. Faculty-supervised projects: AI applications proposed by any Purdue faculty and staff that are supervised by them.
  3. Independent projects: AI applications proposed by the student themselves. These projects are supervised by AI Forge's assistant instructors.
  4. Industry-sponsored projects: AI applications proposed by AI Forge's industry partners. These can span multiple modules and are funded. These projects are supervised by AI Forge's assistant instructors.

Projects

We are developing a Retrieval-Augmented Generation (RAG) + Human-in-the-Loop (HITL) system to enhance AI Forge coursework, with planned deployment in Spring 2025. Imagine a platform akin to Piazza where answers are automatically generated by an AI chatbot based on course materials, but also verified by Teaching Assistants (TAs). The core challenge is designing a scalable TA verification process that ensures accuracy while minimizing manual effort.

We seek highly motivated undergraduate students with a strong foundation in machine learning and deep learning, proficiency in PyTorch, and a basic understanding of Large Language Models (LLMs). This project offers hands-on experience in cutting-edge AI research with direct applications to educational technology.

Project Goals (We plan to get multiple students to multiple goals):

  • Develop a RAG-based system to automate responses to student queries using course materials.
  • Design and implement a human-in-the-loop framework for scalable TA verification of AI-generated answers.
  • Deploy the system in a real-world educational setting (AI Forge coursework, Spring 2025).

Responsibilities: 

  • Develop and refine RAG models to generate accurate and contextually relevant responses.
  • Design and implement mechanisms for efficient TA verification of AI outputs.
  • Evaluate system performance using metrics such as accuracy, scalability, and user satisfaction.
  • Present progress and findings at weekly team meetings.
  • Collaborate with mentors and peers to iterate on the system design.
  • Generate a final report at the end of the semester.
  • Submit to the Fall Undergraduate Research Expo 2025, November 18-21, 2025; Posters on Nov. 18 and research talks on Nov. 19

Qualifications:

  • Current BS, BS/ MS, MS student in Computer Science, Data Science, or Artificial Intelligence at Purdue University.
  • Be taking up to 12-credits on other disciplines
  • Strong background in machine learning.
  • Basic understanding of Large Language Models (LLMs), prompting, and their applications.
  • Experience with real-world data analysis, evaluation metrics, and preprocessing pipelines.
  • Proficiency in programming (Python required; C+ + a plus).
  • Excellent written and verbal communication skills for technical documentation and presentations.
  • Ability to work both independently and as part of a collaborative team.

This project develops an AI-powered Q&A system that enhances public understanding of AI legislation. The system will enable users to query proposed and enacted policies, understand their implications, and explore impacted demographics. Key tasks include creating automated summaries of legislative documents and building a RAG (Retrieval-Augmented Generation) system to provide accurate, context-aware responses. For example:

  • Which states have AI policies for K-12 classrooms?
  • What are the key provisions of these policies?

We seek highly motivated undergraduate students with a strong foundation in machine learning and  roficiency in Python. A basic understanding of Large Language Models (LLMs) and prompting techniques is preferred. This project offers an opportunity to contribute to socially impactful research at the intersection of AI, policy, and public engagement.

Project Goals: 

  • Develop a RAG-based Q&A system to answer user queries about AI legislation.
  • Create automated summarization pipelines for legislative documents.
  • Evaluate system performance using metrics such as accuracy, relevance, and user satisfaction.
  • Deploy the system as a public-facing tool to enhance policy transparency.

Responsibilities:

  • Design and implement RAG models to retrieve and generate responses from legislative databases.
  • Develop automated summarization techniques for long-form policy documents.
  • Evaluate the system using quantitative metrics and user studies.
  • Present progress and findings at weekly team meetings.
  • Collaborate with mentors and peers to refine the system and address challenges.
  • Generate a final report at the end of the semester.
  • Submit to the Fall Undergraduate Research Expo 2025, November 18-21, 2025; Posters on Nov. 18 and research talks on Nov. 19

Qualifications:

  • Current BS, BSMS, MS student in Computer Science, Data Science, Artificial Intelligence, or a related field at Purdue University.
  • Strong background in machine learning and/ or natural language processing (NLP).
  • Proficiency in Python and experience with ML/ NLP libraries.
  • Familiarity with Large Language Models (LLMs) and prompting techniques (preferred).
  • Experience with real-world data analysis, preprocessing, and evaluation metrics.
  • Excellent written and verbal communication skills for technical documentation and presentations.
  • Ability to work both independently and as part of a collaborative team.

This project will develop a framework for tracking how climate change messages evolve on two very different social media ecosystems:

  • Meta's paid-ad environment- climate-related campaigns retrieved through the Meta Ad Library API (e.g., Facebook & Instagram sponsored posts).
  • Bluesky's public timeline- organic posts on a decentralized, AT-protocol network.

The goal is to address the following research questions:

  • What themes, frames, and argumentative strategies dominate climate change advertising on Meta?
  • How do those themes compare with grassroots discourse on Bluesky?
  • How do both spaces change over time (e.g., before/ after major climate events or policy milestones/ changes)?
  • Does the decentralized nature of Bluesky foster measurably different narratives or sentiment compared to the paid, highly-targeted environment on Meta?

Project Goals: 

  • Develop and implement scalable pipelines for collecting and preprocessing climate-related content from Meta and Bluesky.
  • Develop the framework for answering the research questions.
  • Present research results at weekly meetings.
  • Generate a final report at the end of the semester.
  • Submit to the Fall Undergraduate Research Expo 2025, November 18-21, 2025; Posters on Nov. 18 and research talks on Nov. 19

Responsibilities:

  • Design and implement RAG models to retrieve and generate responses from legislative databases.
  • Develop automated summarization techniques for long-form policy documents.
  • Evaluate the system using quantitative metrics and user studies.
  • Present progress and findings at weekly team meetings.
  • Collaborate with mentors and peers to refine the system and address challenges.
  • Generate a final report at the end of the semester.
  • Submit to the Fall Undergraduate Research Expo 2025, November 18-21, 2025; Posters on Nov. 18 and research talks on Nov. 19
  • Based on the project outcome, we will write a paper and submit it to the NLP/ CSS/ AI conference for publication.

Qualifications:

  • Current BS or BSMS student in CS at Purdue.
  • Strong background in machine learning.
  • Ideally, some experience with NLP and transformer-based models (e.g., BERT, LLaMA, or RoBERTa).
  • Solid experience with real-world data analysis.
  • Excellent programming skills (e.g., Python).
  • Ability to work independently and as part of a team.
Causal discovery aims to uncover cause-and-effect relationships from observational data, but traditional methods often struggle with limited data or complex real-world scenarios. This project explores how Generative AI and Large Language Models (LLMs) can enhance causal discovery by incorporating external world knowledge into the process. For example, LLMs can provide contextual insights (e.g., "Smoking causes cancer") to guide causal inference algorithms, improving their accuracy and robustness.

We are developing a novel framework that integrates LLMs with data-driven causal discovery methods to leverage both structured data and unstructured textual knowledge. The goal is to create a system that can identify causal relationships more effectively in domains where traditional methods fall short.

We seek highly motivated undergraduate students with a strong foundation in machine learning and deep learning, proficiency in PyTorch, a solid foundation in statistics, and a basic understanding of LLMs. This project offers an opportunity to contribute to cutting-edge research at the intersection of causality and generative AI.

Project Goals:

  • Develop a hybrid framework that integrates LLMs with data-driven causal discovery algorithms.
  • Design methods to extract and encode world knowledge from textual sources into causal models.
  • Evaluate the performance of the proposed framework on benchmark datasets and real-world applications.
  • Publish findings in top-tier machine learning conferences or journals.

Responsibilities: 

  • Modify and experiment with state-of-the-art causal discovery algorithms (e.g., NOTEARS, DAGMA, DAGPA) using PyTorch.
  • Develop techniques to incorporate LLM-generated knowledge into causal inference pipelines.
  • Evaluate the framework?s effectiveness using metrics such as causal graph accuracy and robustness.
  • Present progress and findings at weekly team meetings.
  • Generate a final report.
  • Collaborate with mentors and peers to refine methodologies and address research challenges.
  • Generate a final report at the end of the semester.
  • Submit to the Fall Undergraduate Research Expo 2025, November 18-21, 2025; Posters on Nov. 18 and research talks on Nov. 19

Qualifications: 

  • Current BS, BSMS, or MS student in Computer Science, Data Science, Artificial Intelligence, or a related field at Purdue University.
  • Strong background in machine learning and deep learning, with hands-on experience in PyTorch.
  • Basic understanding of Large Language Models (LLMs), API use, and prompting (desirable).
  • Ideally, some experience with causal inference methods or graph-based models (desirable but not necessary).
  • Solid experience with real-world data analysis and preprocessing pipelines.
  • Solid foundation in statistics.
  • Proficiency in programming (Python required; C+ + a plus).
  • Excellent written and verbal communication skills for technical documentation and presentations.
  • Ability to work both independently and as part of a collaborative team.

Traditional logistics optimization struggles with real-world complexity, and a key barrier to applying advanced AI is the lack of accessible, high-fidelity simulation platforms. This project tackles this dual challenge by designing and building a new open-source logistics simulator from the ground up. This platform, enhanced with a LLM assistant, will allow users to define complex what-if scenarios using natural language and will serve as a realistic environment to train and validate Neural Algorithmic Reasoning (NAR) models.

Our work involves the co-development of two core components: a modular, discrete-event logistics simulator and a novel AI framework that learns from it. Students will be directly involved in the architectural design and feature implementation of the simulator itself, including the integration of an LLM to ensure the platform serves not only as a testing ground for algorithms but also as a pedagogical and exploratory tool for uncovering insights that traditional methods might miss.

We seek highly motivated undergraduate students with strong software engineering skills and a solid foundation in machine learning and optimization. Proficiency in Python and object-oriented programming is essential for contributing to the simulator's development, while experience with a deep learning framework like PyTorch is needed for the AI modeling. This project offers a unique dual opportunity to build a significant open-source tool and conduct cutting-edge research at the intersection of AI, simulation, and supply chain management.

Project Goals: 

  • Develop an open-source, modular logistics simulator for mid-mile operations from order generation to order fulfillment.
  • Design a neural algorithmic reasoning framework integrating AI, the logistics simulator, and optimization models to control dispatch and management of vehicles, containers, and orders efficiently.
  • Evaluate the performance of the AI-driven framework against traditional optimization methods on benchmark logistics scenarios.

Responsibilities:

  • Modify, experiment, and improve with the SimPy-based discrete-event logistics simulator.
  • Develop and implement Neural Algorithmic Reasoning and Graph Neural Network models using PyTorch.
  • Design and run simulation experiments to evaluate the effectiveness and robustness of the AI framework.
  • Present progress and findings at weekly team meetings.
  • Collaborate with mentors and peers to refine methodologies and address research challenges.
  • Generate a final report at the end of the semester.
  • Submit to the Fall Undergraduate Research Expo 2025, November 18-21, 2025; Posters on Nov. 18 and research talks on Nov. 19

Qualifications:

  • Current BS, BSMS, or MS student in Computer Science, Data Science, Artificial Intelligence, or a related field at Purdue
    University.
  • Strong background in machine learning and deep learning, with hands-on experience in PyTorch.
  • Proficiency in Python programming and experience with object-oriented design principles.
  • Solid foundation in probability and optimization algorithms.
  • Experience with discrete-event simulation (e.g. SimPy) is highly desirable. Familiarity with Graph Neural Networks (GNNs) or algorithmic reasoning is a plus.
  • Basic understanding of Large Language Models (LLMs), API use, and prompting (desirable).
  • Excellent written and verbal communication skills for technical documentation and presentations.
  • Ability to work both independently and as part of a collaborative team.

Despite the growing integration of Generative AI (GenAI) in U.S. K-12 curricula, existing tools fail to address foundational GenAI literacy gaps. Students remain vulnerable to challenges such as hallucination, bias, and ineffective prompting due to a lack of gamified, hands-on learning environments. This project merges constructivist learning theory with agent-based AI to develop an open-source framework where an LLM agent dynamically scaffolds prompt engineering skills within Minecraft. By leveraging tool calling to manipulate the Minecraft environment (e.g., generating blocks, transporting players, spawning entities), the agent enables embodied learning, concretizing abstract GenAI concepts in a familiar gaming context.

You will design, implement, and evaluate an LLM-powered agent that monitors Minecraft server chats, interprets player requests, can do Q&As with players, and performs tool calls to assist players in achieving their goals. Unlike traditional "chatbot tutors" (e.g., Khanmigo), this agent goes beyond conversational support by physically altering the game environment, fostering deeper understanding of GenAI capabilities and limitations. Your work will include:

  • Developing an open-source LLM agent that educates players on effective GenAI querying.
  • Implementing extra adaptive scaffolding mechanisms (e.g., math, geography, and history questions) to provide differentiated support based on player proficiency on subject matter.

Conducting a controlled user study comparing the agent?s effectiveness against static tutorials using metrics such as learning gain (pre/post-tests) and prompt quality (rubric-scored).

Project Goals:

  • Develop an Open-Source LLM Agent for Minecraft:
    • Enable the agent to interpret player requests in natural language and execute tool calls via the Minecraft Java API.
    • Ensure the agent can follow well-structured instructions, teaching players effective prompt engineering.
  • Integrate Gamefied Q&A
    • Gamify domain-specific Q&A for literacy in STEM, geography, and history. Quality of a player's answers can determine the level of agent assistance in the game.
  • Ensure Safety and Ethical Alignment for K-12 Use:
    • Implement safety guardrails (e.g., filtering harmful requests using constitutional AI principles and APIs like Perspective API).

Responsibilities:

  • Design Educational GenAI Experiences:
    • Collaborate with mentors to create engaging, pedagogically sound interactions within Minecraft.
  • Implement Agent Logic and Safety Modules:
    • Use LangChain for tool calling with the Minecraft Java API.
    • Develop a RAG (Retrieval-Augmented Generation)-based system for Q&A.
    • Integrate safety modules to ensure age-appropriate, ethical AI behavior.
  • Modify and Extend Existing Agents:
    • Adapt an existing agent framework to meet project goals, ensuring scalability and modularity.
  • Conduct User Studies and Evaluate Outcomes:
    • Design and execute experiments to measure learning gains and prompt quality.
    • Analyze data and draw actionable conclusions for iterative improvements.
  • Communicate Progress and Results:
    • Present findings at weekly team meetings.
    • Prepare a final report and submit the project to the Fall Undergraduate Research Expo 2025 (November 18-21, 2025).

Qualifications:

  • Current BS, BSMS, or MS student in Computer Science, Data Science, Artificial Intelligence, or a related field at Purdue University.
  • Strong foundation in programming.
  • Familiarity with Linux environments, API integration, and object-oriented programming in Python.
  • Prior experience with Minecraft playing and LLM prompt.
  • Ownership of a licensed copy of Minecraft Java Edition 1.21+ with an active account.
  • Ability to work independently and collaboratively in a fast-paced research environment.