簡單介紹，在Linux環境下進行操作，大概有5大題，但只需要寫2題，寫完2提剩下的時間可以用來照後續問題的資料，作答時間結束後會進來詢問，整個面談時間大約1個小時。
面試的主管會有兩位，把大概的問題問完，就請人離開了，不會跟HR面試，因為會有二面，但我沒被邀請參加估計是被刷掉了
當時面試的考題：
1. Segmentation using segment-anything
2. CV edge detection using Canny
3. 他請我略過 Docker Integration
4. 專有名詞講解
a. What is the difference between "multi-threading" and "multi-processing"
b. Describe what is "UART" , "SDK" , "EVB" , "OSD" , "NPU" , "edge" in embedded systems.
5. Comprehension Questions
a. What is an embedded system, and how does it differ from a general-purpose computer(PC)
b. Explain the process of using AI models for inference, dividing the computation into "pre-process" , "inference" , and "post-process" stages.
c. What do each of these stages represent in terms of computation steps, and how would you allocate these computations between the CPU and NPU?
d. Please provide an example using object detection.
