While AI for Advanced Driver Assistance Systems (ADAS) is growing with a velocity which has never been seen before, there are different challenges that Data scientists have to face in their daily business as there is nothing like “Panacea” solution to every problem. “Time spent” and “Go-To-Market” is more important than ever before and hence, need to optimize and unify AI processes and workflows.
- There is a need of fully-managed solution which provides seamless and unified experience across multiple-cloud offerings and on-premises AI HW/SW and simulation stack. Scaling ML/DL projects from prototype to production and then to deployment should be the integral part of AI pipeline, that requires proper planning and industry proven architectures.
- In order to replicate, share and collaborate ML/DL projects with others, it’s extremely important to monitor and keep track of all the ML / DL experiments with associated data. This can be painful and frustrating, if not done properly. Also, data needs to be consolidated before performing analysis and training. It can be scattered across multiple sites. One should be able to access data wherever and whenever needed, whether in cloud or on premises.
The purpose of this talk is to explore the strategies and architectures that can help us to make AI for ADAS easy to consume and at the same time managing the ML/DL workflow across hybrid cloud platform with little/no effort