Skip to content
View alohays's full-sized avatar
37 followers · 14 following

Achievements

Achievement: Pair Extraordinairex4Achievement: Pull Sharkx3Achievement: QuickdrawAchievement: YOLOAchievement: Starstruckx2

Achievements

Achievement: Pair Extraordinairex4Achievement: Pull Sharkx3Achievement: QuickdrawAchievement: YOLOAchievement: Starstruckx2

Block or report alohays

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.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. 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
alohays/README.md

Hi there, I'm Yunsung πŸ‘‹

AI Research Engineer who bridges research and production

Head of Research, WoRV @ MaumAI Β· Adjunct Professor @ DGIST

I'm an AI research engineer with strong foundations in both research and practical engineering:

  • πŸ”¬ Research background: 10+ publications in top-tier conferences (NeurIPS, CVPR, ICLR, AAAI) with 1.9K+ citations
  • πŸ€– Embodied AI: Leading research on foundation models that unify language, vision, and action for robotics and autonomous driving
  • πŸ’» Engineering experience: Previously shipped commercial AI products at Wrtn reaching 5M+ MAUs
  • πŸŽ“ Teaching: Undergraduate coursework in Physical AI at DGIST, School of Undergraduate Studies
  • 🍸 Fun fact: I'm a licensed home bartender in my free time β€” Korea's national Craftsman Bartender license!

What I'm passionate about

The work I find most interesting happens where a research result meets a running system. My focus right now:

  • 🦾 Vision-Language-Action models and robotics foundation models
  • 🌏 World models for robotics and autonomous driving
  • 🧠 Autonomous agents with memory and personalization
  • πŸ”„ Diffusion models and generative AI
  • πŸš€ Scaling research prototypes to production-ready systems

Let's connect! β˜•

I love meeting curious minds and sharing ideas. Whether you're a fellow researcher, student, entrepreneur, or just someone interested in AI - I'd be happy to chat!

πŸ“… Book a virtual coffee chat with me - 30 minutes of free-form conversation about AI, research, career paths, or anything that sparks your interest.

🌐 More about me: alohays.github.io · CV

πŸ’Ό We're hiring! The WoRV team is looking for people to work on Physical AI and robotics - full-time roles, internships, and μ „λ¬Έμ—°κ΅¬μš”μ› (alternative military service) positions. β†’ recruitment.worv.maum.ai

⭐ Check out some of my open-source work:

Robotics & Embodied AI - worv-ai

  • D2E - Scaling vision-action pretraining on desktop data for transfer to Embodied AI (ICLR 2026)
  • CostNav - A navigation benchmark for real-world economic-cost evaluation of Physical AI agents
  • vla-evaluation-harness - One framework to evaluate any VLA model on any robot simulation benchmark

Desktop agents - open-world-agents

  • open-world-agents - Everything you need to build a state-of-the-art foundation multimodal desktop agent, end-to-end
  • desktop-env - A real-time, high-frequency, real-world desktop environment for desktop-based ML development

Earlier work

Pinned Loading

  1. allenai/vla-evaluation-harness allenai/vla-evaluation-harness Public

    One framework to evaluate any VLA model on any robot simulation benchmark.

    Python 523 41

  2. awesome-visual-representation-learning-with-transformers awesome-visual-representation-learning-with-transformers Public

    Awesome Transformers (self-attention) in Computer Vision

    271 36

  3. open-world-agents/open-world-agents open-world-agents/open-world-agents Public

    Everything you need to build state-of-the-art foundation multimodal desktop agent, end-to-end.

    Python 43 11

  4. worv-ai/CostNav worv-ai/CostNav Public

    CostNav: A Navigation Benchmark for Real-World Economic-Cost Evaluation of Physical AI Agents

    Python 22 1

  5. paper2pr paper2pr Public

    AI/ML Paper β†’ Presentation-ready Beamer + Quarto slides via Claude Code multi-agent workflow

    HTML 5 1

  6. dotfiles dotfiles Public

    Standard-first, agent-friendly dotfiles for macOS/Linux/SSH: one-command install, clean replacement, token-efficient defaults, optional TUI polish.

    Shell 1