Data Science Students: Forget the Models, Focus on Deployment

(This article is a transcript in English of one of my french linkedIn posts)

Recently, I started posting regularly on LinkedIn. As a result, many Data Science students have contacted me, seeking advice on landing an internship, apprenticeship, or first job. The resumes I've seen are often very interesting, but they all have one thing in common: they focus on theory and model understanding. Machine Learning, Deep Learning, using Transformers, Computer Vision models... all of this is great! But I've noticed there is often not much on the deployment side.

The reality is that many training programs don't include teaching how to put models into production. Yet today, the ability to deploy them and manage their life cycle (MLOps) is arguably a more sought-after skill by companies than mastery of the Data Scientist's "classic" toolbox.

If I have one piece of advice to give you: stand out by carrying out one or more projects focused on deployment. Develop web applications with Flask or FastAPI, get familiar with CI/CD, dive into infrastructure as code, and explore tools like SageMaker.

With a little resourcefulness you won't have to pay a cent using Amazon's Free Tier offering.