Developers should visualize code to bring ideas to reality: Sunil Motwani, MathWorks

Tech like AI, big data, cloud computing, IoT, etc. are transforming product design and testing. Sunil Motwani, Industry Director at MathWorks, explains what the company brings to the table for enterprises.    

Sunil_Motwani-edit.jpg

Today, more and more developers, engineers and data scientists are migrating to platforms like cloud to build apps around artificial intelligence (AI), data science, predictive analytics, IoT, among others. 
 
This requires developers to use techniques that had been available to mathematicians earlier to develop complex algorithms. "There are mega-trends changing our world such as automated driving, big data, cloud computing, computer vision, AI, deep learning, internet of things, and software in everything,” says Sunil Motwani, Industry Director at MathWorks.

These market trends are also impacting the way scientists and engineers design and test their products. This is where MathWorks comes into the picture – to create tools that align to market trends. “We make it easy for people to implement their ideas to reality," he adds.  

Why simulation tools are important for advanced engineering

When it comes to training an AI model for an autonomous vehicle, one would not want to build or test a system by sending it on roads with other vehicles and pedestrians. This is where simulation tools can be valuable because they let engineers develop simulated usage in safe environments. This allows them to quickly create better algorithms digitally, which can be tested in the real world. 

...There are mega-trends changing our world such as automated driving, big data, cloud computing, computer vision, AI, deep learning, internet of things, and software in everything.
Sunil Motwani
Industry Director at MathWorks

MathWorks’ Simulink is one such tool helping software engineers test and design products, according to Mike Agostini, Senior Manager, Application Engineering, MathWorks.

According to Agostini, to be successful one must combine AI model building with scientific and engineering insights.

"There are many ways in which AI can fail. AI/ML basically needs lots of data in order to train, as there is a lot of trial and error involved in training such models. We must use tool chains that span the entire design workflow and design how the systems will integrate and interact with the environment,"  Agostini explains.

How do you model AI algorithms for different platforms?

While modelling algorithms for innovative applications may look shiny on the outside, it is incredibly challenging to achieve. Engineers, developers and data scientists need different programming tools where they can implement and test algorithms on different kinds of platforms and devices. For instance, a microcontroller on an edge device, GPUs, or multiple cloud platforms such as AWS and Azure. This is critical so that companies can go to market quickly with their innovation.

...There are many ways in which AI can fail. AI/ML basically needs lots of data in order to train, as there is a lot of trial and error involved in training such models.
Mike Agostini
Senior Manager, Application Engineering, MathWorks

Talking about MathWorks, Sunil Motwani explains how engineers and scientists are leveraging it, “When you model algorithms using our tools, we provide different code generation technologies where you can implement those algorithms on different kinds of platforms.” 

For example, research organizations are using MathWorks’ tools to test ideas - type few lines of MATLAB code to visualize their ideas, and this is helping them innovate on the algorithm. “Our tools can be used at different levels of design,” according to Motwani.

Edited By : Mansi Joshi