A couple of days ago, we had a meeting to help a client operationalise their machine learning model.

About half way into the meeting, they asked:

❓   “How fast do you think you can put our model into production?”

Our response:

💬   “That depends on how well the prototype is built.”

Data science projects often involve a whole bunch of back-and-forth: data exploration, hypothesis testing, changing requirements. All resulting in a great number of data versions and model iterations. Unfortunately and all too often, once the hyperparameters are tuned or visualisations are decided, the team has a very short timeline to ship it to the end-user.

That’s why our team at Melio is so passionate about MLOps and building by starting with the end in mind.

“Today’s prototype is tomorrow’s production app. Build Things Properly.”