Once the data mesh framework became an apparent solution, the BlaBlaCar data leaders began planning a more formalized rollout, starting with a carefully plotted proof of concept. I could see how that related to some of the problems we were facing, and it became pretty clear that data mesh was a very interesting approach-and we were probably heading there without knowing it.” “For me, it was the first time that the concepts that exist in engineering-service-oriented architecture, having microservices, and so on-were expressed so clearly and applied to a data structure. “All of a sudden it clicked into place,” said Emmanuel. The team began partnering data scientists and backend engineers to work closely together, forming what Emmanuel describes as “embryonic squads.” Then, Emmanuel encountered Zhamak Deghani’s landmark article on data mesh framework, a type of platform architecture that leverages a domain-driven, self-serve design. “When you operate that way, structurally, whenever you want to do a project, you require alignment between three to five groups-and that’s super hard to operate,” said Emmanuel. They maintained a team of data analysts, a team of data engineers, and a team of data scientists. Up to this point, the BlaBlaCar data organization was designed around skillsets. That’s essentially what triggered the discussion of ‘We need to do things differently.’” “We had some new use cases around data science that were hard to fit into our existing org. “Data quality that cost us hundreds of hours a year just to investigate, fix, and diagnose was no longer possible,” said Emmanuel. They maintain a modern data stack built on Google Cloud Platform, and as Emmanuel described it, “We have a very strong belief that data-informed decisions create the most long-term value.” Kineret agreed, telling us, “We don’t make any decision without consulting data.”īut in 20, BlaBlaCar took on two large M&As (mergers and acquisitions) that challenged their data team’s speed and quality of work. Image courtesy of BlaBlaCar.īlaBlaCar has been a data-driven company for a long time. Discovering the data meshīlaBlaCar’s data stack was built to support their end goal of supporting self service analytics at scale. To make that happen at scale, the data leaders at BlaBlaCar have adopted a modern data stack and a data mesh platform architecture. The data team even has a mission statement: to deliver dependable data and algorithms to the company. They also build and productionalize algorithms that automate decision-making. The data organization at BlaBlaCar makes sure data flows accurately to consumers, product managers, operations teams, marketing teams, and customer support. The company started as a core app that connected drivers and passengers for carpooling, but today, they support multiple modes of transportation-and immense volumes of complex, peer-to-peer, geographic data. Here’s their story.īlaBlaCar is the largest transportation marketplace in Europe and Latin America for ride-sharing. I recently sat down with three of BlaBlaCar’s data leaders to learn from their journey: vice-president of data Emmanuel Martin-Chave, data analytics engineering manager Kineret Kimhi, and senior data engineer Tushar Bhasin. What about both? Enter BlaBlaCar, which has made real progress on tough challenges like self service analytics and data mesh over the past few years, leading to a reduction in data incidents and time to insights. As part of my job, I’m fortunate enough to speak with data leaders far and wide about how they are tackling some of our industry’s biggest challenges, from implementing data mesh to scaling self-service analytics.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |