Hello. I disagree with points from Rachel and Martin on outsourcing DS.
Indeed, we are the proof that DS is outsourceable and provide it as-a-Service (please se by yourself https://bit.ly/wedo_dsaas). In essence, we've been just doing what the great achievers of Gartner MQ for DSML strategizing today - which is hiring a whole army of data scientists, doing the DS job on behalf of their clients. This despite they initially focused and thought only tooling would be enough.
We can elaborate further on this and would be happy to succeed every single collaboration with our DSaaS team. Please feel free to get in touch