Machine learning algorithm / Data Science
Team currently 4 people , It is expected to expand to 12 people .

One side is technical :
Technical questions :
Take a project introduction .
Service deployment environment ?
GRU and LSTM difference ? Do you try many methods for the same project ?
How is feature extraction done ?
Time domain characteristics 16 individual , Say something ?
feature selection ?
PCA LDA difference ?
PCA principle ?
PCA How to select the most important features .
Division of tasks with other algorithm colleagues .
Youmu has the result according to the algorithm model , In turn, it is suggested that the sensor installation position at the front-end site ? Or for example, it is given above 10 Measuring points , The algorithm will eventually give one 2 Measuring points are the most effective , Just keep this 2 individual .

Introduce to you CV The second item on .
introduce MIP.
PID principle .
LR Where to use it .
lstm Where to use it .

take over C Whether it can be or not? . How long will it take? .
How to connect the data to the front-end system :webapi/mqtt
Have you used it CV/ NLP In the project ? Has convolution calculation been done ?
SVM principle ?
Platform used ? Why is it used tensorflow?
Have you been exposed pytorch
It is called to load the model for reasoning tensorflow Library for ? yes

The docking of algorithm program and main program has special DSP engineer ( Embedded ).

The interviewee is a subsidiary of Lianying medical , Specialized in hearing aids and wearable devices .
Three aspects of algorithm design :
1) Lightweight reasoning on devices , Categorical regression , Decision tree
2) User mobile phone ,TWSR Through wall detection radar
3) User portrait
Signal processing , Machine learning work .

More stringent conditions for engineering deployment
data type :
audio frequency , sensor ,TWSR, infra-red , Motion sensor , Non audio sensor ,IMU( Inertial measurement unit ). User portrait feature label .