Start ordering with Uber Eats

Order now

Enabling Collaboration through Open Source: Highlights from Uber Open Summit Sofia 2019

August 12, 2019 / Global

Uber held its first open source summit on November 15, 2018 in San Francisco. Following the success of this event, we held a second edition of the summit in Sofia, Bulgaria in April 2019. Sofia, home to one of Uber’s first distributed engineering sites, is well-known for its vibrant, growing tech and open source community. 

The Uber Engineering team in Sofia develops systems that enable core business transactions, provide intelligence to power global business decisions, and keep Uber operating in compliance with tax regulations around the world. The team’s culture of transparency and collaboration made Sofia the perfect fit for the second Uber Open Summit.

At the Uber Open Summit Sofia 2019, technologists building and using some of Uber’s most popular open source projects delivered four tech talks. The event also provided a forum for open source community members and developers to engage in discussions about these topics. The summit showcased how open source technologies are driving the future of data visualization, artificial intelligence, site reliability, and more. 

The one-day event consisted of presentations, including practical tips on how to apply open source projects such as M3, Ludwig, Marmaray, and our data visualization suite, including Uber’s Piero Molino, Celina Ward, Lezhi Li, and Eric Sayle shared valuable lessons they’ve learned from working with these projects in a production environment on a daily basis.

The following selection of videos offers an overview of Uber Open Summit Sofia 2019. Be sure to check out the Uber Engineering YouTube channel for more from this and other events.

Keynote: Enabling Collaboration through Open Source

Brian Hsieh and Marin Dimitrov gave the opening keynote for the Uber Open Summit Sofia. They discussed the history of open source at Uber, which began in 2012 and continues today, counting popular and award-winning projects such as, Jaeger, and Horovod. They also explained how Uber’s open source projects provide value to our business and the tech community at large. During the keynote, Brian cited Uber’s 350-plus open source projects on Github and their 2,000-plus contributing engineers from around the world. 

Marin introduced the history and mission of the Uber Engineering team in Sofia. He discussed how contributing to open source helps engineers in a global company mentor each other and build relationships across distributed offices. Marin also explained how open source  helps the team find new learning opportunities and collaborate with other tech experts from around the world.

Ludwig: An open source toolbox for training deep learning models

The summit’s technical presentations began with an introduction to Ludwig, an open source deep learning model training toolbox Uber released in February 2018. During this talk, Piero Molino, who created, maintains, and contributes to Ludwig, provided examples of how to use it. He also dove into Ludwig’s internals with  examples of how to extend it and visualize model performance.

M3: Metrics at Uber

M3 is Uber’s open source metrics stack. At the core of M3 is M3DB, our distributed time series database. In this presentation, Celina Ward walked the audience through how M3 came to be, what it’s used for, and what its future looks like going forward.

Marmaray: A generic, scalable, and pluggable Apache Hadoop data ingestion and dispersal framework

Marmaray is Uber’s open source, general-purpose Apache Hadoop data ingestion and dispersal framework and library. Built and designed by Uber’s Hadoop Platform team, Marmaray allows Apache Hadoop developers to support data ingestion from any source and disperse to any sink using Apache Spark. The name “Marmaray” comes from a tunnel in Turkey connecting Europe and Asia. Similarly, we envisioned Marmaray within Uber as a pipeline connecting data from any source to any sink depending on customer preference. In his talk, Eric Sayle gave audience members strategies for leveraging Marmaray within their own Big Data ecosystems. 

Data Vis: Frameworks at Uber

Uber’s Data Visualization team has developed a suite of open source visualization frameworks dedicated to a range of applications including mapping (,, autonomous driving (, and machine learning (Manifold). During this talk, Lezhi Li described the technology involved in these frameworks and illuminated how they help uncover insights in data for their end users.

Contributing to Open Source: Taking the First Step

Marin Dimitrov led a panel discussion with three software engineers from Uber Sofia’s engineering team (Marina Ilieva, Lyubomir Raykov and Evgeni Kolev) and Vasil Kolev, a member of the Sofia technical community.During this session, the panel members talked about the challenges of getting started with open source and strategies for joining the community. They considered motivating factors, such as improving coding skills, and discussed how being involved  with open source projects has helped them grow in their careers.  

Despite starting their open source journey in different ways, our panelists all agreed that such collaborative initiatives provide good opportunities for learning and improving their skill sets, as well as building long-term relationships that can help them accelerate their professional development. 

Want to get involved? Uber Engineering Sofia is hiring. If  interested, apply here. To attend future Uber Engineering events, join our Meetup group.

Subscribe to our newsletter to keep up with the latest innovations from Uber Engineering.