On any given night, 2 million people stay in homes on Airbnb in 81,000 cities all over the world. There are nearly 6 million listings in 191 countries to choose from —that’s more than the top five hotel chains combined.
Â
What makes all of that possible? Trust.
Â
At Airbnb we see Trust as the fundamental currency of the sharing economy, and it is the heart of everything we do.
Join us for our next Tech Talk where you'll hear how our Trust team detects, deters and takes actions against fraudulent acts to ensure the safety of our host and guest community.
An RSVP is required to attend. You will receive a confirmation email if you register correctly. All proceeds, less fees, will be donated to Women Who Code, the largest and most active community for technical women in the world. Their award-winning programs support over 167,000 members around the globe to level up in their technical careers, every step of the way. They produce 5,000 trainings, workshops, hackathons, conferences and more as well as give away $1.6M in technical scholarships and industry conference awards. Â
Since launching in 2009, Airbnb has grown to a global scale and has become one of the biggest names in the alternative accommodations industry. As the first double-sided hospitality marketplace operating in 191 countries, we are faced with a tremendous responsibility to keep our community safe. Trust is the foundation of our double-sided marketplace - without it, neither guest nor the host can confidently interact on our platform.Â
In this talk, we will discuss what we’ve learned in scaling Trust globally and how we’ve adapted in fighting fraud.
Airbnb is an online marketplace for travel where guests can book experiences and homes listed on the platform by hosts. The ability to control both sides of a potentially large dollar-amount transactions as well as the online and offline nature of our business makes us particularly susceptible to sophisticated attackers. In this talk, we will share the nature of fraud and abuse we face, describe the unique challenges associated with them and provide a glimpse into some of the machine learning techniques we are using to stay one step ahead of attackers.
Machine learning is a powerful tool in detecting fraud at scale. However, as of today every machine learning model requires a significant amount of human effort to deliver reliable results in production. Majority of that time is dedicated to implementing new and maintaining existing features. In this talk we will share how our feature management system enables better online and offline consistency, point in time correctness and order of magnitude faster exploration on new signals.
Due to the high value of transactions that take place on the platform, Airbnb is often an easy target for various types of fraud. To detect and mitigate different vectors of fraud in real time, we've built a highly scalable and performant system known as the Enforcement Framework, which evaluates riskiness of all relevant requests and mitigates the attempted fraud based on our Trust and Safety policies. In this talk, we'll share underlying goals and abstractions that we used while building these systems. We'll then cover in more detail the core component of our platform - the rules engine.
Â
Â
Â
When fraudsters target Airbnb, the best case is that all of our automated systems perfectly detect and remove all malicious activity with 100% certainty. Sometimes our systems need some help and humans have to get involved to do manual investigations and provide labeling to help train our systems. The Trust Tools team maintains a UI platform that supports robust and extensible tooling for other Trust teams to employ against their respective fraud vectors. We leverage a variety of tools to be able to move fast with consistency and confidence: Yeoman/npm run scripts, React and TypeScript, and an abstraction layer over data access. We also built a question and answer state machine for labeling and application of policies. This talk will further explain the motivations and tradeoffs in building this platform.
Airbnb has evolved from a community of individual hosts and guests to a world consists of travel agencies, boutique hotels and property management companies. To adapt to the business needs, we need to modernize our infrastructure to support the relationships between individuals, teams and organizations. This helps us to build trust by reducing account sharing and misrepresentations. In this talk we will review our challenges, learnings and newfound opportunities in supporting businesses as first-class citizens of Airbnb.
