Machine Learning Engineer

About the Company

Worldcoin is a new, collectively owned global currency that will be distributed fairly to as many people as possible. Worldcoin will launch by giving a free share to everyone on Earth. We believe that this is an essential step to accelerate the transition towards a more inclusive global economy, providing new ways for everyone to share future prosperity. We hope you’ll join us on our ambitious journey.

About the Team:

For Worldcoin to launch successfully on a global scale, we need to both ensure fairness and build trust with our users. Therefore, it is essential to prevent fraud by making sure no one can sign up multiple times and erode trust in the system.

The high level goal of security at Worldcoin is to analyze threats and develop measures to detect and prevent them. We consider many threats ranging from on-device security over presentation attacks to backend security. The cross-disciplinary nature of this team requires interfacing with various other teams across the company including Economics, AI, Infrastructure and IoT.

About the Opportunity:

As a Machine Learning Engineer on Worldcoin's Security Team, you will be involved in deploying our hardware device to unsecured environments while maintaining the overall integrity of the system. In this role, you will be responsible for advancing the state-of-the-art fraud detection system on our device. On a high level, some of the questions you will be facing are:

• Which features of a human are reliable to detect and hard to fake?
• Is the object in front of the device a real human?
• How can we effectively mitigate fraud while preserving the user’s privacy?

Since Worldcoin can only achieve global coverage with sufficient security in place you directly impact the success of the project. In this role, you will work closely with different teams to collect data as well as design and implement fraud detection algorithms. These will be iteratively deployed in field tests, as well as attacked by internal and external red teams. Specifically, this role includes tackling problems like:

• Implement vision based fraud detection algorithms
• Extract features from video and sensor streams
• Train robust models and attack them yourself
• Deploy algorithms to the field and monitor them
• Think about further robust fraud detection algorithms & test hypotheses
• Pass on data collection & defense implementation requirements to IoT team 
• If desired and appropriate your responsibility can be increased over time and include project management tasks

About You:

• Someone who thrives in a fast-paced environment that is continuously evolving
• Deeply care about what you are working on
• Own problems end-to-end, and are willing to pick up whatever context is needed to get the job done.
• Have worked in the the following fields:
• Computer Vision
• Machine Learning
• Deep Learning (mostly CNN)
• Sensor Fusion
• 3+ years of programming experience (e.g. Python, PyTorch / Tensorflow, Rust)
• Experienced interacting with Mongo and AWS
• Experienced with monitoring deployed models
• Experienced with edge computing devices
• Work with exceptional care and the ability to understand code in depth.
• Able to read and understand scientific papers, reproduce results, and transfer techniques to other domains
• Strong foundation in probability / statistics
• Are deeply curious and creative
• Are constantly thinking about edge cases
• Bonus: Experience working on KYC fraud detection and other security-critical systems