Skip to content
View oney-erge's full-sized avatar

Block or report oney-erge

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
oney-erge/README.md

Oney Erge

Agentic AI + Physics | UT Austin PhD

Portfolio, projects, publications, and patents: oneyerge.com

I work on language-model and agentic systems for complex, real-world problems.

My background spans data science, machine learning, optimization, robotics, physics-based modeling, and statistics. Much of my research has focused on combining physics-based and data-driven models, especially in fluid dynamics, automated operations, and autonomous systems.

I am especially interested in work that brings AI, robotics, and physical systems together.

Focus Areas

  • Large language models and efficient inference
  • Agentic and multi-agent systems
  • Robotics, simulation, and embodied AI
  • Scientific machine learning and physics-based modeling
  • Optimization, statistical modeling, and autonomous systems

Projects

Project What it does
YBM An assistant for your own computer that runs on a local language model and records what it did. Tested on a 9B model and an 8 GB laptop GPU, with the results and the misses published.
Afterimage A research toolbox for running and studying LLMs on constrained hardware, with memory, placement, and inference strategies plus reproducible benchmarking.
LocalDeploy A hardware-aware platform for finding, deploying, managing, and benchmarking local AI models across different runtimes.
AgentGymnasium A visual physics environment where LLM agents build, simulate, evaluate, and iteratively improve physical designs.
Creature Lab Shove a real robot, see exactly why it fell, and prove the fix. Published robots and their trained policies, in a browser app on a laptop CPU.
SegCraft A config-first toolkit for semantic segmentation training, evaluation, and image/video inference.

Selected Background

  • PhD in Dynamic Systems and Control from The University of Texas at Austin
  • Work across language-model applications, NLP, optimization, and MLOps
  • Research in hybrid physics-based and data-driven modeling
  • 10+ years as Associate Editor, overseeing peer review and editorial decisions for 250+ scientific manuscripts

Links

Oney Erge on LinkedIn Oney Erge on Google Scholar Oney Erge on ResearchGate

Outside of AI: I grew up on 90s and early 2000s PC games and still go back to them. Some things just hold up.

Pinned Loading

  1. YBM YBM Public

    An assistant for your own computer that runs on a local language model, asks before it acts, and records what it did.

    Python

  2. Afterimage Afterimage Public

    A research toolbox for running and studying LLMs on constrained hardware, with memory, placement, inference, and benchmarking methods.

    Python 1

  3. LocalDeploy LocalDeploy Public

    Find, deploy, manage, and benchmark local AI models for your hardware across Ollama, llama.cpp, and compatible runtimes.

    Python

  4. SegCraft-Semantic-Segmentation SegCraft-Semantic-Segmentation Public

    A config-first toolkit for semantic segmentation training, evaluation, and image/video inference.

    Python 2 2

  5. AgentGymnasium AgentGymnasium Public

    A visual physics environment where LLM agents build, simulate, evaluate, and iteratively improve physical designs.

    Python

  6. Creature-Lab Creature-Lab Public

    Shove a real robot, see exactly why it fell, and prove the fix. A browser app that runs published robots (Microduck, Unitree G1) with their real trained policies in MuJoCo, on a laptop CPU.

    Python