CPU Inference Benchmark
PyTorch, ONNX Runtime and OpenVINO on one laptop CPU, FP32 vs INT8, with a controlled measurement method.
- 3.48×ResNet-18 INT8 vs FP32, OpenVINO
- 2.6 msOpenVINO INT8, batch 1
Mukhammaddiyor Abdulazimov
AI student at Harbin Engineering University · ML inference & full-stack
3rd-year Artificial Intelligence student (English-taught) at Harbin Engineering University. I benchmark and quantize models for CPU inference, build reproducible ML evaluations, and have 2+ years of full-stack experience on real client projects. Open to AI/ML and full-stack internships.
Featured projects
All projectsSST Baselines
Six baselines, a small LSTM, spatial-context and whole-region (U-Net, ConvLSTM) models, evaluated reproducibly on daily sea surface temperature (NOAA OISST) in two North Pacific regions.
- 6baselines compared
- 107tests
RoadAI
Transfer-operations platform (MVP, pilot in preparation): driver app, dispatcher dashboard and LLM document extraction with human review.
- 646automated checks
- 3test suites
Gesture Control
Controls the computer with hand gestures through a webcam, using MediaPipe hand tracking and OpenCV.
- 29.9 FPScamera + tracking, CPU
- 228tests in ~1.6 s