Open to remote AI/ML internships

Qiuwei Liu

AI Systems / ML Engineering

Building reproducible AI systems across GPU scheduling, agent workflows and computer vision.

GPU Scheduling · Agent Workflows · Computer Vision

Predict → Schedule → Execute

GPU Scheduling

Plan → Review → Continue

Agent Workflows

Detect → Track → Analyze

Computer Vision

Featured Projects

More Projects

Agent tooling

ChatGPT Web Bridge

Attach-only CDP bridge with skill / MCP integration and explicit execution boundaries.

GitHub
YOLO detection demo

YOLO Model Optimization

YOLO Model Optimization Portfolio

Python · PyTorch · YOLOv8 · Error Analysis

mAP500.238 → 0.315+32%
same frozen validation split (8-class TACO)
  • dataset audit & frozen evaluation split
  • FP / FN and per-class diagnosis
  • small-object bottleneck analysis
  • controlled optimization & standardized evaluation

75.5% of validation objects were Tiny Objects — analysis identified small-object detection as primary bottleneck and motivated higher-resolution experiments.

Diagnostic TP/FP/FN statistics are used for failure analysis and may differ from COCO-style AP evaluation.

Skills

AI Systems / GPU

GPU SchedulingDiscrete-Event SimulationModel CachingCP-SAT

Machine Learning

PyTorchPredictionComputer VisionYOLOU-NetModel EvaluationExperimental Design

Agents / Backend

FastAPIREST APIRTSPAgent WorkflowsMCP / CDPProject Continuity

Engineering

PythonGit / GitHubLinuxCondaPytestCUDAApple Silicon MPSReproducible Experiments

About

South China University of Technology — Master's Student in Mechanical Engineering, 2025 – Present. B.Eng. in Vehicle Engineering.

Focus: AI Systems · GPU Scheduling · Agent Workflows · Computer Vision

Let's build something useful.

Open to remote AI/ML internships and technical collaboration.

Email qiuweiliu03@gmail.com