Journey to scale | Step 3 | Data Scientists

Step 1 | The Hero Use Case

Step 2 | Understanding Operations

‍

It might sound obvious, but in order to have AI/ML running at lab scale, you need someone to build the algorithms and models from the data you have. We usually find that clients have at least one data scientist allocated to building predictive or regression models, typically for BI use cases. In some circumstances the data scientists are also tasked with exploring and experimenting with ML tools and platforms to see if they can augment their internal capability (stay tuned for more on this next time).

If the initiatives are only about integrating off-the-shelf tools and models, we find that either the data scientists struggle with the software complexity of deployment, or deployment and integration are not even attempted. This results in either a POC that is only good for a demo or solutions that are not appropriate in terms of cost, performance or technical debt.

Data scientists should be used to explore the data, select features, train models, test models and prepare the work for the production step. They should not just be working in SQL or spreadsheets!

If you don't have any data scientists, fear not, Arcanum offers professional services to help kickstart your AI/ML journey.

Next, Step 4: ML Experiments

‍