Yun Hong

Yun Hong (洪运)

I am a B.S. candidate in Computer Science at Fudan University and a former exchange student at The University of Texas at Austin.

Currently, I am a Test Development Engineer Intern at ByteDance (Douyin R&D, Live Streaming & Media Stream), working on LLM-powered QA efficiency tools.

Email / CV / Google Scholar / GitHub / ORCID / Kaggle

Research Interests

I am interested in embodied intelligence, robotics, and machine learning — particularly world models for embodied agents and their role in action evaluation, future prediction, and planning under uncertainty. I also work on imitation learning, visuomotor policy learning, and vision-language-action models.

Education

Fudan University Sep 2023 – Present	B.S. Candidate in Computer Science, Shanghai, China Completed risk-map research for autonomous driving at MagicLab under Wenchao Ding; the work is accepted by IEEE Robotics and Automation Letters (RA-L).
UT Austin Aug 2025 – Dec 2025	Exchange Student in Computer Science, Austin, Texas, USA Completed multi-task active perception research for robotic information gathering under Junhong Xu; contributed to benchmark design and policy evaluation.

Research Experience

RA-L 2025	Learning Risk Map for Autonomous Driving in Partially Observable Environments Yun Hong, et al. IEEE Robotics and Automation Letters (RA-L), 2025 paper Engineered risk field representations with spatiotemporal modeling. Developed realistic traffic-scene generation using diffusion models and gradient optimization. Designed a lightweight neural network for efficient risk prediction.
UT Austin Sep – Dec 2025	Multi-Task Active Perception for Robotic Information Gathering Research Assistant, advised by Junhong Xu Built multi-task partially observable simulation benchmarks with unknown object properties and spatial locations. Implemented and compared imitation learning, reinforcement learning, and IL-pretrain + RL-finetune paradigms on fine-tuned VLAs. Reviewed prior work on manipulation exploration and long-horizon POMDP benchmarks.
Fudan NLP Sep 2024 – Jun 2025	ChatGPT-o1 Reproduction and Reflection Ability Improvement Research Assistant, advised by Prof. Xuanjing Huang Constructed, evaluated, and optimized reflection datasets to improve model reasoning. Synthesized self-critic data based on Llama3.1-8b-instruct and generated diverse error-to-correct reflection trajectories. Applied tree-search and reverse-reasoning strategies; evaluated on GSM8K.

Internship

ByteDance
Jan 2026 – Present

Test Development Engineer Intern, Douyin R&D, Live Streaming & Media Stream, Shanghai
Refactored and expanded internal inspection and test cases, improving accuracy from 89% to 100% and coverage from 95% to 98%. Developing an LLM-powered efficiency tool for automated QA workflows.

Projects

LCDP-Sim Dec 2025 – Present	Language-Conditioned Diffusion Policy for Robotic Control Built an end-to-end Vision-Language-Action system using diffusion policy. Integrated CLIP encoder and U-Net DDPM/DDIM architecture for high-fidelity action generation. Implemented 16-step action chunking with a full ManiSkill2 data-train-eval pipeline.
MapNavigation Dec 2024 – Present	Route Planning and Navigation System Combined OpenStreetMap with Gaode Maps API for route planning and location search. Implemented hybrid point selection, detailed route display, and multi-road-type support with speed constraints.
NeuralStyle Jun 2025 – Present	Modular Neural Style Transfer Toolkit Developed a modular Python toolkit for neural style transfer with strong maintainability. Built batch processing and a configurable training pipeline for reproducible experiments. Provided an interactive web interface for in-browser stylization.

Skills

Machine Learning	Supervised & unsupervised learning, reinforcement learning, imitation learning, model evaluation and generalization
Deep Learning	CNN/RNN/LSTM, diffusion models, spatiotemporal modeling, PyTorch-based model design, training, and fine-tuning
LLM & Vision-Language	LLaMA series fine-tuning, VLA models, reasoning and reflection mechanism design
Robotics & Perception	Active perception, multi-task POMDP, information gathering, long-horizon manipulation
Autonomous Driving	Motion planning, trajectory prediction, decision-making, risk-field modeling
Programming	C, C++, Python, LaTeX
Languages	Chinese (native), English (fluent)

Honors & Awards

International	2024: Finalist, Water, sanitation, and hygiene for the prevention and care of neglected tropical diseases (Geneva, Switzerland)
Domestic	2025: Outstanding Work Award, The 5th Meituan Business Analysis Elite Competition 2024: 3rd Prize, China Undergraduate Mathematical Contest in Modeling (CUMCM) 2023: Finalist, Full-stack AI development engineer skills training by NVIDIA

Website template adapted from Jon Barron.