Astitwa Saxena

I have experience in building machine learning models which are efficient and can work on resource constrained devices. My interest in building systems which are able to solve all and every problem which can be solved using computers is what drives me.

The most complex project I have worked on is creating a Lip Reading model which works for mobile devices. Reducing the number of parameters for the state of the art model for visual speech recognition was indeed a challenging problem which I was able to solve. My team brought down the size for the model a massive 120mb size to 6mb while trading off as little as 6 percent in accuracy.