Re-imagining DS @ CARS24
Data science suffers from perspective, we are trying to revisit it via our ‘data platform’.
At first glance, we are just another vertical in an organisation that is trying to help business become more efficient. We can also relook it from a larger perspective and say we are just another piece in a puzzle. But, when we zoom in, we realise we are another puzzle altogether.
What are we doing?
We are moving away from monolithic kind of DS services. When we say kind of, it means that our approach is not opinionated but evolutionary. As more use cases come in, we will enable different ML system designs.
Our workflows as described in the previous set of blogs, were more around solving a problem and were developed keeping that at the center. Few years later, things have evolved and to cater to this need, we are building something you can call a Data Kitchen, that will serve DS/ML solutions at scale.
Why are we doing this?
There is no independent solution in DS, it is always an ensemble of different ideas put together in a way to solve business requirements. We @C24 are continuously growing capabilities and building answers that can be cross leveraged across, at scale.
You can imagine every coloured block in the diagram as a cuisine specific desk in our data kitchen. For every business order we get, we can serve the order by connecting different desks and building on capabilities of our data platform folks e.g. a car damage assessment model using images can be consumed by different business solutions. These endpoints need not be the primary reason for the development, but if enabled correctly can give incremental gain across as represented in diagram below.
How are we doing this?
We are starting a periodic series of blogs where different desks will share something from their work and how that is enabling a scalable and smarter decision making. So stay tuned ! :)
Contributors : Swapnesh Khare , Senior ML engineer @ CARS24; Abhay Kansal , Staff Data Scientist @ CARS24
