The world’s leading publication for data science, AI, and ML professionals.

How Can I Write Better Articles on AI?

I'm pretty transparent about the fact that my writing on medium gets me interesting connections that can turn into leads in my business…

I’m pretty transparent about the fact that my writing on medium gets me interesting connections that can turn into leads in my business development pipeline. The analytics for the 35 articles I have published on medium.com tell me that you – my readers – like to read more about the business of AI, and my experience as a consultant. You don’t seem to care too much about technical advice or learning resources.

The data are telling me that articles like "How to Hire an AI Consultant" and "How to Price an AI Project" are 10 times more interesting to you than a demo on how to extract email validity, or a demo on machine vision.

The Pivot: Less Science More Vision

Well, rather than fight the data, I’m going to listen to your clicks and claps, and focus my articles on the experiences I’ve collected. It will take some extra cognitive effort, but I’m going to do this without violating any NDAs, basically by storytelling with illustrative examples, instead of using declarative "I did X for company Y." In fact, I’m promising in this post not to be a name dropper. I’ll give you my honest take on what’s going on from my perch on a branch overlooking the maelstrom of AI engineering and commercialization. So no more "here is the code to do X." Also, from a strictly selfish point of view, writing the code for those articles was not trivial. And so when I write about viewpoints and experience, it costs me less time and makes me more eyeballs.

I asked myself when starting this series of articles, how I can express the things my company does, without violating NDAs. These are things like regression, classification, normalization, cloud, IoT, education, and so on. Instead I’m pivoting to talk about experiences and viewpoints. Like my business partner Matt points out, my efforts need to make money, save money, or save work and time. This pivot will save work and time (my work and time), and it will make money (by introducing you, my audience, to me). And so, dear reader, if I have misinterpreted your lust for tales of business in the AI world, and all you really wanted was technical details…. tell me so. Post a comment.

Experiences and Viewpoints: Billable Data Cleanup

And now, on to my first of many experiences to write about… A client recently asked me to help them organize their data before they would commit to a contract. What this attitude misses is that data science is all about data, not algorithms. The nice graphic below gives you a sense of just how huge a job it is to clean up your data. That usually is a bigger chunk than the whole rest of the project.

From Mohamadreza Mohtat in this post.
From Mohamadreza Mohtat in this post.

Now, once the data is organized, it takes some time to fit a solution to the data, and to hook the solution into the architecture. Usually our client is going from a database of some sort (e.g. postgres) into tensors, and then back from tensors into some sort of callable JSON service that pumps out a response.

So to wrap this up, when I do AI consulting, I don’t let the data cleanup escape the statement of work. It IS the work. The rest of the project may seem like the hard part to the client, but that’s not the hard part for the consultant. The hard part is getting your data into shape. The rest is magic.

If you enjoyed this article then please try out the new clap tool at the bottom right of your screen. Like I said earlier, I’m also happy to hear your feedback in the comments.

Happy Coding!

-Daniel [email protected] LemaySolutions.com


Related Articles