Project Pinkly-Blue
A cross-disciplinary exploration of dawn-themed interface design.
View Case Study
Day Dreamer, Hamburger Eater, and Garbage Creator;
Busy doing nothing, accomplishing even less.
Hi there! This is Ziqi ZHAO, and you can also call me
Stephen. Currently, I am a year-4 student at the School of
Data Science, the Chinese University of Hong
Kong, Shenzhen (CUHKSZ), majoring in Data Science and Big
Data Technology.
My research interests lie in time series data forecasting,
financial data analysis, and AI applications in cloud-native governance fields.
Outside of research, I am an enthusiastic fan of the San Antonio
Spurs and Manchester City. I also have a deep passion for music,
especially Rap and K-pop, with Jony J and
IU being my all-time favorite artists.
Find the story beyond the data. User's journey is the soul of every project I touch.
Dare to reinvent the ordinary, shattering abstract constraints to build bold, original digital experiences.
Perpetually in beta...A lifelong voyager, navigating and evolving within the ever-shifting trends.
Exploring the workplace as a product manager and trying my best to create value bravely and strongly.
A student at the School of Data Science, majoring in Data Science and Big Data Technology, ranking top 20%.
I had a very happy, fulfilling and meaningful childhood at Ningbo, a very beautiful coastal city in the east part of Zhejiang Province.
In an era of information overload, ColorMood serves as a visual diary that bridges the gap between abstract feelings and aesthetic clarity. With its unique "One Photo, One Palette" philosophy, it transforms everyday randomness into a curated museum of personal moods, helping users find their own "base color" and connect with those who resonate on the same frequency.
TOGETHER WITH Yasi YANG*, Yixuan ZHOU, Wanru YAN in course
IDE2000
PurrCircle is a hyper-local pet care platform that solves
the "travel anxiety" of 80 million urban pet owners in
China. By activating community trust through a dual-track
model, it connects owners with verified neighborhood helpers
for affordable, in-home care.
TOGETHER WITH Zhiyuan YANG, Shengxi Xiong
A next-generation authentication system that replaces
traditional methods with real-time hand gesture recognition.
Powered by a custom CNN with Residual and SE blocks, the
system achieved 99.7% accuracy across 13,000+ self-made
samples, successfully bridging high-performance Computer
Vision with frictionless user cybersecurity.
Xinxin Liu, Ruixiang Mei,
Ziqi Zhao
, Shaohan Wang, Junli Duan*
This paper introduces a high-precision dual-model framework
to predict smart port operation efficiency, addressing the
challenges of nonlinear dynamics and noisy multi-scale data.
By integrating CEEMDAN for signal decomposition with a
lightweight Echo State Network (ESN) and a Bidirectional
Attention-Enhanced LSTM (Bi-ALSTM), the system achieves a
significant leap in temporal modeling. Experimental results
demonstrate a 30.5% reduction in MSE compared to traditional
RNNs, while maintaining a real-time training speed of just
0.35 seconds, providing a robust quantitative tool for
dynamic scheduling and risk early warning in global supply
chains.