BIGDATA-HACK Hackathon in Moscow

On December 9-10, Gett and McKinsey hosted the BIGDATA-HACK hackathon at McKinsey’s offices in Moscow.

The hackathon started with 3 keynote lectures by Gett:
George Nichkov presented a review of the industry Gett operates in.
Evgeni Begelfor presented Gett’s solution to supply repositioning.
Ianir Ideses presented Gett’s approach to implementing reinforcement learning in its products.

George Nichkov

 

Evgeni Begelfor

Ianir Ideses

McKCrowd

24 teams participated in the hackathon, each team selected its preferred solution based on its core competencies. Tracks included textual analysis, GIS problems and marketplace dynamics. The participants were also free to tackle any problem they wished that could be solved by utilizing the data supplied by Gett. Consequently, there was a great variety of solutions that were presented to the jury.

The participants were given 24 hours to analyze the data, think of an interesting problem with real world value and build a solution.
To support this spurt of creativity, participants were given data, technical assistance, mentoring, and all the main food groups (in the form of burgers and pizza).

HackathonPizza

After 24 hours, each team was given 10 minutes to present their solution to the technical jury. In these 10 minutes, the participants presented the problem they aimed to solve, their solution and their code.

ReviewingHacks

Having reviewed the participants’ solutions, the jury selected 12 teams that proceeded to the finals. All teams, finalists or not, were given feedback for their work, finalists were given guidelines on how to present their work in front of the full jury.

Jury Panel

In the final stage, each team was given 10 minutes to present to the jury. Points were given for the technical level of the solution, for the task complexity and for the presentation itself.
It was a very close race, the level of work that was shown by the contestants was very high. Nonetheless, 3 teams shone above the rest:

3rd place winner:
Team Kanape introduced an algorithm to fit the distribution of drivers to riders. In addition, the team showed an NLP algorithm to extract crucial information from user reviews.
kanapepres

 

KanapePodium

2nd place winner:
Team Getto presented methods for supply redistribution based on driver and rider heatmaps.

gettopres

GettoPodium

1st place winner:
Team MacDacHack presented a method to improve rider pickup by identifying complex rider locations (malls, metro stations, etc’). Once a complex location is identified, a smart pickup location is selected based on rider destination and driver location.

macdachackpres1macdachackpres2MacDacHackPodium

Judging from the participants’ feedback, the hackathon was a great success, stay tuned for a similar event in Tel-Aviv.

TheGang

What people are saying about this article

Be the first one to comment.