After 6 months working at kumparan, these are my lessons learned (mostly non-technical stuff). I hope it will helps you to make decision on your career (if any) 🤝.

A little bit background first, there are 3 projects that I lead so far (in ordered timeline):

1. Kumparan’s Tracking Events Delivery System Services
2. Kumparan’s Trending Services
3. Multi-Universe Kumparan’s A/B Test Platform

Mostly on data platform stuff, this is to make sure that kumparan’s data team is working on its full potential. There are other projects such as topic/content recommender and content analyzer stuff that i’m part of.

You may notice that my first project is a tracker system. There are no trending service or any data services that require users data can be built without the tracker. So, that’s why my first task in kumparan is to design and implement tracking events delivery system at kumparan’s scale.

I learned to always implement the simplest and complete solution first. A few months ago, I have implemented bayesian-based Google News trends algorithm using SQL-only on Google BigQuery. No complex implementation. I’m not suprised that Google AI team is  using SQL to implement gradient descent and integrate it to Google BigQuery. BigQuery is very powerful tool!

I learned to work smart. Time is very limited, I tend to all write all the details upfront before writing any code. It ‘s helps me to know tools or library that I need and helps me estimate the timeline. Writing the details on upfront saves me a lot of time and boost my productivity.

I learned to write a good documentation. A documentation is not just for other people, but it’s for me too. I tend to forgot the details in 1–2 months after the project is running on production 😅. The documentation will helps me to get the context of the previous project in minutes. I’m able to write good documentation based on the feedback from my team, so all credits goes to my team 🙇‍.

I learned to communicate well and still practicing. My biggest struggle is explaining my ideas to others. A simple rule that I learned from a great project managers at kumparan is: context first. But in practice, it’s very hard to do. I tend to assume that other people already on the same page as me. I will keep improving my communication skill.

My biggest regret is using Google BigTable as part of kumparan tracker arch. It’s so expensive af 😓. But it scales very well, if I use self-hosted redis/cassandra at this scale then maybe I can’t sleep 🤣.

This list is subject to change and thanks for reading!