Ethan Haarer

HELLO, WORLD!

package contactCard;

public class Ethan extends Student {
    public Ethan() {
      String fullName = "Ethan Haarer";
      String school = "Georgia Tech";
      Masters major = "Computer Science";
      String specialization = "Machine Learning";
      Resume resume = new Resume("pdf");
      String[] hobbiesAndInterests = new String[] {
        "Hockey",
        "Reading",
        "3D Printing",
      }
    }
    public static void contact() { }
}

EDUCATION

Georgia Institute of Technology

B.S. - Computer Science

Focuses in Intelligence and Devices

Graduated Summa Cum Laude May 2024

Georgia Institute of Technology

M.S. - Computer Science

Specialization in Machine Learning

Expected Graduation December 2025



PROFESSIONAL EXPIRENCE

AI Software Engineer Intern

L3 Harris

May 2025 - August 2025

  • Worked within the Space and Airborne Systems Division under the Intel and Cyber sector on the L3 AI Concept Incubator (LACI) team to support cross-sector internal research and development opportunities on emerging LLM and ML technologies.
  • Prototyped LHX Intelligence (LHXi) for a L3 Harris-common voice controlled intelligent AI assistant that supports ISR missions at the edge to reduce operator cognitive load.
  • Developed LLM-based multi-agentic framework that organizes and schedules task topologies to objectively evaluate reliability and improve explainability & trust in products integrating LLMs and statistical ML Models.
  • Demonstrated integration and capabilities of ML and extensible LLM technologies into CI/CD workflow to shareholders at Q2 Demo to communicate how new advances are reliable and scalable, improving confidence in customers.
  • Experimented with MCP integrations to evaluate integration techniques on varying model sizes and capabilities, and orchestrated dockerized solutions for ideation handoff.



  • Graduate Research Assistant

    Teachable AI Lab @ Georgia Tech

    August 2024 - Present

  • Working alongside Zekun Wang on TAIL's Cobweb Project, an unsupervised clustering model which supports incremental concept learning from fewer examples. The model learns both structure and parameters where the structural expansion is dynamic. It uses multimodal approach which is human-like as it is able to develop and update new concepts within the web structure
  • Implemented a novel look-ahead mechanism for the Cobweb algorithm to optimize hierarchical clustering as measured by enhanced flexibility and depth control by introducing a BFS-inspired approach with tunable frontier layers.
  • Conducting experiments to benchmark Cobweb against traditional neural networks and other clustering models. This involves assessing its ability to learn from fewer examples, its adaptability in dynamic environments, and its performance in comparison to other state-of-the-art models in terms of accuracy, scalability, and efficiency
  • Working on implementing a supervised learning variation to cobweb by implmenting Neural Net techniques like backtracking, alongside hyperparameter search using the Weights & Biases API.



  • EDS AI Data Engineering Intern

    KPMG US

    June - August 2024

  • Integrated generative AI into software development lifecycle, integrating automatic optimizationand standardization into developers' workflow that saw a near 30% increase in developers' efficeny through streamlining development and vastly improving code readability
  • Oversaw optimization of FOT pipeline utilizing new optimization techniques and implemented major time complexity improvements in 30% of the codebase across thousands of lines of code form O(n2) to O(n log n) leading to a near 20% reduction in pipeline runtime
  • Researched potential modes to further integrate unit generation into SDLC to prioritize expanding code coverage of generated tests to streamline codebase improvement overhaul by integrating a complete end to end pipeline to automate SparkSQL to pySPARK conversion
  • Analyzed impact of Generative Ai in developpment of code and pipelines to determine the inherent benefit of services like chat GPT, CoPilot, and Databricks Ai in optimizing code and reducing overall company costs



  • Enterprise Data Engineering Intern

    Homesteader's Life Company

    May - August 2022

  • Developed and implemented an IT Asset Management system to reduce missing and orphaned assets, maximized useful life and ROI, providing greater insights into asset utilization, capacity management, maintenance expense, and asset depreciation
  • Reduced mean repair time by approximately 20% through Asset management system which maintatined accurate REST API to facilitate cross-organizational active updates to inventory and tracking, leading to an increase in ROI for over 1,300 assets
  • Developed data framework and integrated a new tracking system to manage asset ownership and forensic backlogging for IT and all departments; utilized Power BI and Postman



  • Peer Instructor

    The Hive @ Georgia Tech

    January 2022 - May 2024

  • Assist and instruct hundreds of electrical engineers in prototyping and configuring projects from ideation to engineering, fabrication, testing and deployment. Frequently focused on rapid integration into exisitng project goals and accomplishments to aid end user's in project development, troubleshooting and ad hoc solutions
  • Led and facilitated hands-on workshops for over 150 electrical engineering students, focusing on advanced circuit design, PCB fabrication, and prototyping techniques. These workshops not only enhanced the practical skill sets of the participants but also contributed to a 25% increase in project completion rates, as students were better equipped to translate theoretical concepts into tangible prototypes.
  • Optimized the use and maintenance of lab resources, resulting in a 15% reduction in equipment downtime and a 10% decrease in material waste, ensuring sustainable operation of the makerspace.
  • Certified in technical toolsincluding general EE tools such as waveform generators and oscilloscopes, PCB fabrication, 3D printing, Laser Cutting, and Soldering
  • FEATURED PROJECTS

    Filter Buttons

    Outpost - The Wearable LLM Agent on the Edge

    Outpost is a wearable voice-enabled assistant powered by a quantized LLAMA model and utilizes RAG to provide information about how to survive outdoors! In addition, it hosts a YOLO model that identifies animals with the onboard camera, and sensors to warn of environmental dangers.
    Here's the Demo Video
    Uses a Raspberry Pi 5, and Ai Camera, and framework built in Python

    Reggie - Gatech LLM Registration Chatbot

    Chatbot utilizing LLM combined with RAG and live API-fed data to get registration logistical information, class slot openings and professor reviews for Georgia Tech
    Click here to check it out
    Development Documentation
    Github Repo
    Built in Python using HuggingFace

    Analyzing Cross Cultural Typicality in Language Models

    A study that examines whether a multilingual large language model (GPT-4o) aligns with human typicality norms across five languages and mirrors human cognition.
    Link to the Study
    Github Repo
    Used Python to collect and analyze GPT-4o dataset

    Predicting Pedestrian Positions with YOLOv5 and Gen Ai

    Researched different methods to predict people's position in a top-down view with accurate distance estimation from POV images
    Full Paper Linked Here
    Github Repo
    Written in Python with Google Collab

    Inspiration AI - DnD Generative AI

    Website to help DnD enthusiasts generate new locations, characters, quests, and more utilizing DnD REST API, LLMs, and stable diffusion models
    Alpha Build Website
    Github Repo
    Built in Python using Streamlit

    Cobweb 3 LLM Optimization

    Working within TAIL to optimize Cobweb, a clustering model which supports incremental concept learning from fewer examples. The model learns both structure and parameters where the structural expansion is dynamic.
    Link to Lab Page
    Github Repo
    Built in C++ optimized with CUDA

    QUENCH - Robotic Bartending Machine

    Prototpye drink mixer that decodes RFID card data to pour the perfect drink. Also hosts a native web server to request custom drinks over WiFi!
    Demo Video
    Find the Code Here
    Written in C using Arduino

    Mapping Crimes in Italy Capstone

    Created a website for George Mason University to view and filter crimes in early Italy for historical research. Ability to upload custom maps and datasets
    Link to the Site
    Github Page
    Written using React, Javascript, and Python