To apply for this job, you need to complete both steps below:
STEP 1:
Please click the link to submit your application directly to the company:
Your application will only be received by Recruiter if submitted via above link.
STEP 2:
Kindly scroll to the bottom of this page and complete the short VinUni Tracking Form.
Filling out this form alone does not count as applying. Kindly remind this form is not part of the company's application process. It only helps Careers, Alumni, Industry and Development (CAID) Department discover more opportunities and follow up in case of system issues.
Key Responsibilities
- Deploy CNN/DNN and vision models onto embedded AI platforms using runtimes such as Qualcomm QNN, TensorRT, TIDL, ONNX Runtime, or equivalent.
- Convert and optimize models from PyTorch/TensorFlow → ONNX → target inference runtime.
- Optimize models using FP16, INT8, quantization, PTQ/QAT, operator fusion, and graph optimization.
- Analyze operator compatibility, graph partitioning, accelerator mapping, and CPU fallback.
- Optimize AI workloads across heterogeneous CPU/GPU/DSP/NPU architectures.
- Profile and optimize latency, FPS, throughput, CPU/accelerator utilization, memory footprint, and memory bandwidth.
- Analyze and reduce data-movement overhead through DMA, shared memory, zero-copy, buffer management, and efficient memory pipelines.
- Optimize end-to-end pipelines such as:
- Camera → Pre-processing → AI Inference → Post-processing → Application
- Identify and resolve performance bottlenecks related to compute, memory bandwidth, synchronization, scheduling, and hardware utilization.
- Integrate and debug AI inference pipelines in embedded Linux environments using C/C++ and Python.
Requirements
Required Qualifications
- Bachelor's or Master's degree in Computer Science, Computer Engineering, Electronics, Embedded Systems, AI, or related fields.
- From 3 years of experience in similar positions.
- Strong programming skills in C/C++ and working knowledge of Python.
- Good understanding of CNN/DNN, tensors, computational graphs, neural-network operators, and AI inference.
- Hands-on experience with ONNX and at least one embedded AI inference runtime or accelerator SDK.
- Understanding of FP32/FP16/INT8, quantization, PTQ/QAT, and model optimization.
- Understanding of heterogeneous computing using CPU, GPU, DSP, NPU, or dedicated AI accelerators.
- Good knowledge of embedded system concepts including multithreading, synchronization, memory management, shared memory, and IPC.
- Experience with profiling, performance analysis, and bottleneck identification.
Preferred Qualifications
Experience in one or more of the following is highly desirable:
- Qualcomm QNN / Hexagon DSP / HTP
- NVIDIA CUDA / TensorRT / Jetson
- TI TIDL / C7x / MMA
- Camera and computer-vision pipelines
- OpenCV, OpenVX, GStreamer
- DMA and zero-copy architectures
- DDR/cache/memory-bandwidth optimization
- Real-time or high-performance embedded systems
- Automotive ADAS, robotics, or edge-AI products