With an aim to promote “hackathon” culture at Indix (and also, of course, to get some of those crazy ideas implemented), we set ourselves a goal of conducting one internal Hackathon per quarter. With the hackathon for Q1 having been a huge success, we conducted the one for Q2 on the 25th and 26th of June, 2015.
Bigger and Better
Indix has grown a lot since the previous hackathon in January. We crossed the 50 employees landmark at the Chennai office in May. With more people, the energy and excitement around another hackathon was bigger than ever.
The previous hackathon taught us that we need to get the ideas floating around much ahead of the event, and we were able to advertise the ideas (and their owners) in advance.
The event really turned out to be bigger and better – compared to 12 ideas that were worked on and demoed last time, we got 17 ideas this time around.
Here is a sample of some of the wonderful ideas and their solutions that were demoed.
One of the products is a Chrome plugin called Indix Livewire. It is a simple solution to solve the problem of keeping everyone in the company updated on the various (internal) links that we access everyday. The plugin is a new tab extension, and is backed by a simple Google sheet. The Google sheet enables anyone to easily add and update the links, and is a godsent for new joinees.
At Indix, we are heavy users of GoCD. Our team recently got second place in the GoCD plugin contest for the gocd-s3-artifacts plugin. Continuing in that vein, a team also worked on enabling elastic agents for GoCD with Mesos. This will help us in scaling agents across our multiple GoCD instances based on the demand. And of course, we expect to open source the same after testing it in our infrastructure for a few months.
We also saw come to life an Excel-based visualizer for our API. This aims to get users more accustomed to spreadsheet-based workflows to leverage our amazing API to get the data that they need.
One of the most exciting demos was on extracting attributes from images. The particular implementation focused on apparel, and it enabled extraction of attributes like style, sleeve, color, gender and more. It made for both an interesting demo, and a promising product. Another image-based hack was Visual Search, which when given an image, gave back similar images.
As a data company that does a lot of crawling, we are also looking for different ways to acquire the data. With the recent trend of companies moving towards mobile apps only (or first) strategy, we also had a solution for a mobile applications based crawling system. This particular implementation utilized mobile app automation framework Appium to drive the process. Another team worked on improving the efficiency of our AJAX-based crawls by coming up with a way to detect and allow only the necessary AJAX call(s) from a web page.
Another effort was in coming up with a “page rank” for product pages based on the seller influence, other products that they sell, and other inputs. The page rank will aid in the search and efficiency of re-crawling.
We are also overdue for conducting an external Hackathon. Our Data Monster data hackathon was last conducted in March 2014. With the momentum and excitement gained around our internal hackathon, we are all geared up to conduct another edition of Data Monster soon. Stay tuned for more updates on that!