3 ms·

well, creating metrics to measure product usage, creating dashboards that help summarize and visualize these metrics isn't exactly in the same boat as working o

well, creating metrics to measure product usage, creating dashboards that help summarize and visualize these metrics isn't exactly in the same boat as working on search relevance. Most of the dashboards in my company are created using tools like Tableau or some JS/Python/MySQL. At best I am learning how to hack together some simple web.

ptio 13y ago
Since you are interested in search, see if you can use Kibana for your dashboards (http://www.elasticsearch.org/overview/kibana/ http://www.elasticsearch.org/overview/kibana/). Kibana uses ElasticSearch which uses Lucene. If you are able to convince your team to use Kibana then maybe you can ask your boss to spend some company time contributing to ElasticSearch and/or Lucene.
dkural 13y ago
Tableau is a great tool to discover opportunities to improve search relevance, through visualizing data. Did you come up with any ideas through your work with the metrics pipeline & dashboards? Understanding the metrics intimately, building some intuition for them via dashboards, and plotting things in order to formulate hypotheses is a good way to get you warmed up. It seems that by these projects, they're trying to familiarize you with what they measure. For projects like improving search relevance being able to formulate an insightful theory and quickly checking this against logs/metrics is way more important than programming skills. It seems that you're working with all the right tools - come up with an insight & make a pitch for an A/B test to the right boss!
hamburglar 13y ago
It helps to realize any problem (ok, most problems) can be interesting if you look for ways to make it interesting. I hacked together some simple web for my company's internal dashboards and they work fine but I have a million ideas for how I can improve them, but I've always got more pressing priorities to worry about. As a result, I never actually have the time to spend on them. If my manager told me "I want you to spend the next two weeks working solely on the dashboards" I could interpret it as shit work (I mean, come on, we all realize dashboards are pretty boring) or an opportunity to get creative. My job satisfaction has increased greatly with the ability to drop the ego and take an unglamorous task and just do a good job on it.
- nostrademons 13y ago
  "we all realize dashboards are pretty boring" Hah, one of my first jobs at Google was to do the dashboards for our feature's dogfood, and my manager had to tell us "I know that engineers love to tweak dashboards and make them look pretty, but please try not to spend too much time on them, we have other important things to do."
nostrademons 13y ago
What is it that other people in your company wish the dashboards did that they don't now, and what would have to happen for that to exist? It's really common for early versions of dashboards or metrics to use off-the-shelf components because the person who set them up had other responsibilities (like building the product) and didn't have time for it. That doesn't mean they should stay that way forever. They could be giving you this assignment as a test to see what you do with it - if you replace the off-the-shelf stuff with pipelines and displays that are way better, that shows that you have the skills that'll be necessary for working in IR or ML.

ptio 13y ago

Since you are interested in search, see if you can use Kibana for your dashboards (http://www.elasticsearch.org/overview/kibana/ http://www.elasticsearch.org/overview/kibana/). Kibana uses ElasticSearch which uses Lucene. If you are able to convince your team to use Kibana then maybe you can ask your boss to spend some company time contributing to ElasticSearch and/or Lucene.

dkural 13y ago

Tableau is a great tool to discover opportunities to improve search relevance, through visualizing data. Did you come up with any ideas through your work with the metrics pipeline & dashboards? Understanding the metrics intimately, building some intuition for them via dashboards, and plotting things in order to formulate hypotheses is a good way to get you warmed up. It seems that by these projects, they're trying to familiarize you with what they measure. For projects like improving search relevance being able to formulate an insightful theory and quickly checking this against logs/metrics is way more important than programming skills. It seems that you're working with all the right tools - come up with an insight & make a pitch for an A/B test to the right boss!

hamburglar 13y ago

It helps to realize any problem (ok, most problems) can be interesting if you look for ways to make it interesting. I hacked together some simple web for my company's internal dashboards and they work fine but I have a million ideas for how I can improve them, but I've always got more pressing priorities to worry about. As a result, I never actually have the time to spend on them. If my manager told me "I want you to spend the next two weeks working solely on the dashboards" I could interpret it as shit work (I mean, come on, we all realize dashboards are pretty boring) or an opportunity to get creative. My job satisfaction has increased greatly with the ability to drop the ego and take an unglamorous task and just do a good job on it.

nostrademons 13y ago
"we all realize dashboards are pretty boring" Hah, one of my first jobs at Google was to do the dashboards for our feature's dogfood, and my manager had to tell us "I know that engineers love to tweak dashboards and make them look pretty, but please try not to spend too much time on them, we have other important things to do."

nostrademons 13y ago

What is it that other people in your company wish the dashboards did that they don't now, and what would have to happen for that to exist? It's really common for early versions of dashboards or metrics to use off-the-shelf components because the person who set them up had other responsibilities (like building the product) and didn't have time for it. That doesn't mean they should stay that way forever. They could be giving you this assignment as a test to see what you do with it - if you replace the off-the-shelf stuff with pipelines and displays that are way better, that shows that you have the skills that'll be necessary for working in IR or ML.