Poki Data Scientist

Why Do You want to Work at Poki?

• grown up playing games on websites such as miniclip
• data science experience : interesting challenge
• sounds like a fun place to work
• perfect next step in my career, learn and grow with Poki and be a part of people enjoying games on the web

Growing up, I used to spend countless hours playing games online on websites such as Miniclip (does anyone remember Club Penguin? :) ). Being a data scientist with a master in AI, finding ways to ensure players have a nice experience using Poki while also being able to make sure they find the games they want sounds like a great challenge. Poki sounds like a fun place to work, and this would be the perfect next step in my career.

It is not every day that you find a position that you think you might enjoy and also know that you can contribute to upon being given a chance, and I did not want to let this opportunity slide.

Can You Describe a time where You Identified an Area of Improvement and the way You Followed up on That?

• In my work at Emirates NBD (one of the national banks of Dubai), we were building an analytics pipeline to identify fraud transactions. While the pipeline worked, they were using a Microsoft service to do the analytics and the performance was not great.
• In my time there I helped them build an analytics pipeline from scratch using Python + SQL. I contributed quite a bit to the feature engineering steps that would later be part of the final analytics program used across Emirates NBD.

In my work at Emirates NBD (one of the largest banks in Dubai), the team was attempting to improve an analytics pipeline to identify fraudulent transactions. While the pipeline worked reasonably well, they were using a Microsoft service to do the analytics, and the performance was not great. In my time there, I helped them build an analytics pipeline from scratch using Python + SQL. I contributed quite a bit to the feature engineering steps that would later be part of the final analytics program used across Emirates NBD. Many of those changes led to improving the final accuracy by a decent amount and made it much easier for my coworkers to push the product to the upper management.

When You Look at the Presentation of Content on Our Platform, what Question Comes to Mind?

• Clustered, icons are similar
• Computer vision + analytics: Check for something like that.
• Given an example

While Poki looks great, one thing that immediately comes to mind is that many of the icons are rather visually similar (eg : Apple Knight, Apple Knight mini dungeons ; Mekabolt, Day of Meat). There are also a lot of icons immediately, which is slightly overwhelming. This is unintentional, of course, but might make it harder for a user to pick what they want. Using a computer vision similarity technique could be useful in identifying them and pushing them somewhere else on the grid. This would be something that I would love to discuss as well.