Open to opportunities

QingYang Liu

@qingyangliu

Message

AI researcher specializing in generative models, multimodal perception, and safe embodied-agent systems.

Malaysia

Message

What I'm looking for

I seek research-focused roles where I can build safe, interpretable AI for real-world systems, collaborate across disciplines, publish impactful work, and deploy models under strict latency and safety constraints.

I am a highly motivated AI researcher focused on generative models, federated learning, and multimodal perception with applied experience in autonomous driving and medical imaging.

I have led end-to-end research projects, producing novel methods such as a physics-guided generative planning framework for adverse-weather autonomous driving and enhanced CycleGAN variants for weather-aware scene generation.

My work spans implementation, large-scale experiments, ablation studies, and reproducible releases; I have contributed to multiple accepted and under-review papers and open-source codebases, achieving strong empirical gains on benchmarks and real-world datasets.

I thrive in collaborative, interdisciplinary teams but can drive independent research; I prioritize robustness, interpretability, privacy, and safety when building impactful intelligent systems.

Experience

Work history, roles, and key accomplishments

Perception Algorithm Engineer

Nullmax

Aug 2025 - Oct 2025 (2 months)

Designed and implemented CV/3D perception algorithms and multi-sensor fusion for ADAS, optimized DNNs for ≤30ms on-board latency and ≤8GB memory, and built automated data engines to halve annotation costs.

Sensor Fusion CNN Transformer Optimization TensorRT Model Quantization Active Learning

Research Contributor

Monash University

Mar 2025 - May 2025 (2 months)

Contributed to a VAE-Transformer hybrid for image rain removal that improved PSNR/SSIM on Rain100H and Rain1400 benchmarks and designed the Latent-Guided Attention module.

Transformers Latent Guided Attention Benchmarking PyTorch Ablation Studies

Research Contributor

Monash University

Jan 2025 - Apr 2025 (3 months)

Co-developed a hybrid GAT+LSTM framework for multi-bus trajectory prediction achieving 88.12% short-term and 66.12% long-term mission accuracy on a Kuala Lumpur bus dataset.

Graph Attention Networks Dynamic Graph Fusion Ablation Studies

Research Scientist

Monash University

Nov 2024 - Apr 2025 (5 months)

Led development of WeatherPhysDiff, a physics-guided diffusion planning framework for autonomous driving under adverse weather, reducing average displacement error by 31% in rain and achieving 90% speed-constraint compliance in dense fog.

Diffusion Models Control Barrier Functions Semantic Physical Embeddings NuPlan PyTorch Autonomous Experimental Design

Research Contributor

Monash University

Jan 2024 - Dec 2024 (11 months)

Co-authored a multi-modal masked autoencoder pipeline (Mamba) for 3D brain tumor segmentation, improving BraTS whole-tumor Dice and delivering an efficient, reproducible inference pipeline.

Masked Autoencoders Modality Aware Masking PyTorch Medical Imaging Docker Statistical Analysis

Research Engineer

Monash University

Jun 2024 - Aug 2024 (2 months)

Developed an enhanced CycleGAN for weather-conditioned driving-scene generation that improved YOLOv8 mAP50 from 59.6% to 93.0% on augmented datasets and published reproducible code.

CycleGAN Multi Scale Discriminator SSIM Loss YOLOv8 Model Evaluation

Software Development Engineer

Nanjing MySkysoft Technology Co., Ltd

Jul 2022 - Aug 2022 (1 month)

Led front-end architecture and a four-person team to deliver a mobile-optimized marketing web application with a component library that reduced development cycles by over 30% and increased mobile engagement by 23%.

Component Driven Development Responsive Design State Management Performance Optimization Accessibility

Education

Degrees, certifications, and relevant coursework

Monash University Malaysia

Master of Artificial Intelligence, Artificial Intelligence

2024 - 2025

Grade: GPA: 3.167/4.0 (Weighted Average Mark: 77.5/100)

Activities and societies: Major courses included Deep Learning (87 HD), Multi-agent Systems and Collective Behaviour (80 HD), Fundamentals of Artificial Intelligence (83 HD), and IT Research Methods (83 HD).

Master of Artificial Intelligence program at Monash University Malaysia with coursework in deep learning, multi-agent systems, AI fundamentals and IT research methods, achieving a 3.167/4.0 GPA.

Soochow University

Bachelor of Internet of Things, Internet of Things

2019 - 2023

Grade: Weighted Average Mark: 87.7/100

Activities and societies: Honors/Awards: One-Time Recipient of National-Level Awards; Twelve-Time Recipient of School-Level Awards; Triple-A student Model.

Bachelor degree in the Internet of Things from Soochow University with a weighted average mark of 87.7/100 and multiple institutional honors and awards.