Let's talk about Business Value

I'd like you to grasp the crucial distinction between side projects and the real world: solutions created with cutting-edge deep learning technology for some arbitrary activities will only be relevant in the real world if they provide value - I mean genuine business value. Realising this fact early on in your career enables you as a practitioner to focus your efforts on what matters most for organisations, institutions, and corporations that have actual consumers and clients.

So, lately, I've been working on an intriguing tool that uses Transformers and other AI technologies. Rather than focusing my presentation on how amazing transformers are (even though they ARE) or how I created a dataset to train deep learning models. I've forced myself to consider what business value the clientele I'll be demonstrating this demo tool want out of AI solutions.

Consider, for example, using a sentiment analysis algorithm to analyse client feedback on specific items and take marketing or sales action based on consumer comments about those items. If a product has an abundance of negative reviews, a company's effort would be to decrease the number of stock orders for that product and focus on correcting the cause of the lousy feedback.

My usual approach to presenting the implemented AI solution would have been to address the technical and architectural aspects of transformers and the training procedure for whichever model I employ. Your current or future business clients don't necessarily want to focus on the technical nerdy details. Instead, they have their customers at the forefront of their minds, and so should you and I.

Podcast

With all that's been said above, transformers are very interesting, and the multimodality of transformers is even more intriguing. Although transformers are best suited for natural language processing(NLP) tasks, recently, we've seen the utilisation of transformers for vision-based problems. In my conversation with Lewis Tunstall, co-author of the widely popular book "Natural Language Processing with Transformers" we explore transformers in detail and Lewis's journey within the machine learning industry.

Subscribe to the podcast on YouTube, Spotify and Apple​

​

Gems

If you are interested in expanding your knowledge of Data Engineering, then I would recommend the following books:

• ​Data Pipelines with Apache Airflow By Julian de Ruiter and Bas Harenslak
• ​Fundamentals of Data Engineering By Joe Reis and Matt Housley (Early Release)

Cultivating your data engineering knowledge with both books will equip you with the technical, architectural and fundamental knowledge required by proficient data engineers. Be on the lookout for future data engineering articles coming from me.
​

Thanks for reading!