Don’t be a Machine Learning Hobbyist

Do you know Dave Farley? If you don’t, you better go and subscribe to his YouTube channel. He has amazing videos on software engineering and continuous delivery. He is actually one of the authors of the famous book “Continuous Delivery“.

In one of his videos, titled “Your Programming “Hobby” WON’T Cut It”, the importance of becoming a better software developer through building a product to meet an actual business requirement. He shares that it’s completely fine for a person to play around and develop software as a hobby, but if you want to up your game, you really need to interact with a stakeholder.

I’ve handled a few 100E Projects during my time in AISG. For your information, a 100E project is an industry project scoped in a way that a machine learning product is delivered to clients (or as we call them, “project sponsors”) within a 7-month period. The project sponsors are the stakeholders here, and they expect the machine learning model to work in their use case. They also expect the software to be written in a way that is understandable and useable in the future.

If I were to extend the message from Dave Farley’s video, I would say that this:

You might know how to do data science by listening to courses online, or you might even be able to participate in a Kaggle competition, but…
Until you have a project sponsor providing you with unclean data and demanding requirements for the performance of a model and,
Until you have to hand over high quality, reproducible code to people on a wide spectrum of knowledge regarding machine learning, you aren’t upping your game.

When it comes to development, it’s very important to do exploratory learning via hobbies to learn general matters. However, It’s only when we do professional development based on the requirements of stakeholder that we learn the most and produce high quality code.

That’s what the AIAP program is for.
That’s what a 100E will give you.
It will give you a unique experience and challenge for you to learn the ropes.

The AI Apprenticeship Programme (AIAP)

The clients or project sponsors are putting money into the program for a model to be delivered to them. Also, the apprentices who join the AIAP learn how to deliver a fully working AI software product to the sponsors. It’s a win-win for both the apprentices and the project sponsor.

And it’s not just about modelling. The whole end-to-end of a project will be handed over to the sponsor: from the code used for performing EDA to the files used to generated Docker containers.

The 7-month timeline of a 100E project

At the end of the day, the feeling of accomplishment is super important. The feeling that you’ve managed to solve the problem given to you in the best possible way, and learn along the way. This sense of accomplishment is not an illusion. It’s the product you have delivered. The sponsor also shares the joy of your accomplishment.

Author

  • Siavash Sakhavi

    I love engineering and science. I’ve always been interested about how the world works and how we can make it a better place. ?  I enjoy teaching, guiding people, and sharing my experiences. I’ve had experiences in applying machine learning methods for applications such as audio, point clouds, and texts.

    View all posts