Platform
Engineering Service
Opt-AI provides engineering services and software platforms for optimizing large AI models on resource-constrained edge devices.
This includes model building, model compression, and embedded system optimization technologies
Model Building
-
Hardware-friendly redesigning of AI models for target edge devices
-
Providing hardware and AI model recommendations, performance profiling, and automatic model tuning
Model Compression
-
Reducing model size while improving inference speed and maintaining accuracy
-
Providing quantization, pruning, and knowledge distillation
Embedded System Optimization
-
Porting (C++) and optimizing embedded systems
-
Providing parallel processing techniques for inference scheduling and accelerated computing
On-device AI Software Platform
Opt-AI’s goal is to support customers in securing AI inference performance, reducing development costs, and accelerating market launch by providing ① model building, ② model compression, ③ embedded system optimization technologies, and ④ a HW-aware SW analysis system that can automatically perform optimization in various HW environments for the deployment of on-device AI.
Module
-
Model Builder + Model Compressor + Model Deployer
-
HW-aware SW analysis system
Benefit
-
Securing performance on specific HW + reducing costs + accelerating market launch
-
Minimizing additional development costs due to hardware changes
Core Technology
Opt-AI ensures effective performance on any hardware and processor through our HW-aware SW Profiling System.
We support customers in focusing solely on model development by building development workflows specific to target hardware
Supported Hardware
-
Qualcomm, Ambarella, Renesas, NXP, NVIDIA, Xilinx, Rockchip, Coral
Supported Processor
-
CPU, GPU, NPU, FPGA
Supported Task
-
Classification, Object Detection, Human Pose Estimation, Image Enhancement
Supported Hardware
