Jingyao Gong

Ph.D. Student, Nanjing UniversitySuzhou, Jiangsu, China

I am a Ph.D. student at Nanjing University, advised by Prof. Wenbin Li in the RL-MIND Group. My research focuses on efficient AI models. I am also interested in large language models, including pre-training, post-training and reinforcement learning, and their extension to vision and speech.

I received my M.S. (2025) and B.Eng. (2022) from Jiangsu University, where I worked with Prof. Heping Song on image compressed sensing, and then worked as an algorithm engineer at Westlake Institute for Data Intelligence.

gongjy.cs@foxmail.com

jingyaogong@smail.nju.edu.cn gongjy@stmail.ujs.edu.cn

Google Scholar GitHub Hugging Face ModelScope

Education

Nanjing University

Ph.D. Student (Advisor: Prof. Wenbin Li)Sep. 2026 - Present

Jiangsu University

M.S. in Computer Technology (Rank: 1/187, Advisor: Prof. Heping Song)Sep. 2022 - Jun. 2025

Jiangsu University

B.Eng. in Software Engineering (Rank: 1/61)Sep. 2018 - Jun. 2022

Experience

Westlake Institute for Data Intelligence

Algorithm Engineer, LLM pre-training and agentic RLDec. 2024 - Jul. 2026

Calix

Algorithm Intern, retrieval-augmented code generationDec. 2023 - Nov. 2024

Honors & Awards

Outstanding Master's Thesis, Jiangsu University2025
Outstanding Graduate, Jiangsu University2025
Champion, AI Hackathon, Westlake Institute for Data Intelligence2025
National Scholarship for Graduate Students2024
First Place, National AI Competition, Visual Feature Coding Track2023
Second Prize, "Huawei Cup" National Graduate Math. Modeling Contest2023
First Prize, Jiangsu Graduate Mathematical Modeling Competition2023
Silver Medal, Kaggle Happywhale Competition2022
National Scholarship for Undergraduate Students2022
Outstanding Undergraduate Graduate, Jiangsu University2022
Second Prize, National College Students Mathematics Competition (twice)2019 - 2021
First Prize and Second Prize, Jiangsu College Students Mathematics Competition2019 - 2021
First Prize (Jiangsu), Lanqiao Cup Programming Contest2018 - 2022
First-Class Scholarship (three times); Merit Student (twice), Jiangsu University2018 - 2022

Publications* Student first author

MiniMind-O Technical Report: An Open Small-Scale Speech-Native Omni Model

J. Gong

arXiv preprint arXiv:2605.03937, 2026 Preprint

[Paper] [Code]

Multi-Scale Wavelet Convolutional Sampling and Cross-Stage Feature Enhancement Reconstruction for Image Compressed Sensing

L. Qin, Z. Xiong, J. Gong, Z. Zhong, Y. Yi, H. Song

Pattern Recognition, 2026, accepted Journal

[Preprint]

Representation Sampling and Hybrid Transformer Network for Image Compressed Sensing

H. Song, J. Gong^*, H. Jia, X. Shen, J. Gou, H. Meng, L. Wang

IEEE Transactions on Circuits and Systems for Video Technology, 2026 Journal

[Paper]

Context-Aware Detail Recovery Network for Single Image Super-Resolution

H. Song, D. Zhou, J. Gong, H. Jia, X. Shen, J. Gou

Engineering Applications of Artificial Intelligence, 186: 116476, 2026 Journal

[Paper]

Multi-Scale Sampling and Windowed Sparse Self-Attention Reconstruction Network for Image Compressive Sensing

H. Song, L. Zhang, J. Gong, H. Jia, X. Shen, J. Gou, Y. Lai, H. Meng

Image and Vision Computing, 106181, 2026 Journal

[Paper]

Multi-Cross Sampling and Frequency-Division Reconstruction for Image Compressed Sensing

H. Song, J. Gong^*, H. Meng, Y. Lai

Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), 38(5): 4909-4917, 2024 Conference

[Paper]

SALSA-Net: Explainable Deep Unrolling Networks for Compressed Sensing

H. Song, Q. Ding, J. Gong, H. Meng, Y. Lai

Sensors, 23(11): 5142, 2023 Journal

[Paper]

Back Projection Deep Unrolling Network for Handwritten Text Image Super Resolution

H. Song, H. Ma, Y. Si, J. Gong, H. Meng, Y. Lai

Computers and Electrical Engineering, 111: 108965, 2023 Journal

[Paper]

Projects

jingyaogong/minimind

Training a small language model from scratch, covering tokenizer, pre-training, SFT, LoRA, DPO, PPO/GRPO, tool use, agentic RL and distillation in native PyTorch.

jingyaogong/minimind-v

A small vision-language model trained from scratch with visual alignment and multimodal instruction tuning.

jingyaogong/minimind-o

A ~0.1B speech-native omni model with text, audio and image input and streaming speech output.