Perception Intern; Summer

About Zipline

Zipline is the world’s largest and most experienced drone delivery service. We are on a mission to serve all humans equally by ensuring access to food, medicine and essential goods anytime, anywhere. We design, build, and operate the world’s largest autonomous logistics system, delivering critical supplies quickly and reliably. Today, Zipline operates on four continents, makes a delivery somewhere in the world every 30 seconds, and has completed millions of deliveries to date, including blood, vaccines, medical supplies, food, and retail products.

Our customers include the world’s largest and most prominent healthcare systems, governments, retailers, restaurants and global businesses who rely on us to save lives, reduce emissions, increase economic opportunity, and provide delivery from point A to point B as fast as possible. The drone is only 15% of what we’ve built to enable seamless, reliable, global operations.

Our system strengthens supply chains, reduces congestion, and gives people time back. With more than 140 million commercial autonomous miles safely flown, Zipline is redefining access to healthcare, consumer products, and food across the globe.

We operate at a global scale and are looking for practical problem solvers who thrive on real-world challenges and rapid growth. Our team is motivated by building systems that have a direct, meaningful impact on people’s lives and by scaling the future of logistics. We are seeking people who sculpt from first principles, enjoy facing adversity, and can do the impossible at record breaking speeds.

The Droid Perception team owns both the onboard perception system that operates in a constrained compute regime on the drone and the offboard perception system that is built to augment, validate and improve the onboard system. Both onboard and offboard perception systems are camera-only systems that do not use LiDAR. The team’s work involves:

• Creating robust pipelines to process large datasets to train various models
• Training cutting-edge machine learning models of various types and sizes that understand the 3D structure as well as the semantics and enable us to make safe deliveries at scale
• Writing clean, high-quality code to train models to utilize GPUs efficiently
• Building autonomous systems that run on the drone as well as in the cloud that use the machine learning models to power Zipline’s P2 platform
• Optimizing the onboard models into TensorRT engines to deliver high throughput within a limited compute budget
• Understanding the failure modes of the onboard and offboard perception systems and designing solutions to solve them
• Interfacing with the planning, navigation, maps, and hardware teams to understand each other’s needs and design systems accordingly

As an intern in the Droid Perception team, you will dive into ML model experimentation, evaluation and integration to push the boundaries of our current onboard and offboard perception systems.

Our distinctive challenges require groundbreaking approaches, and you'll have the chance to see your ideas transition from concept to real-world application. In a fast-paced and collaborative environment, you'll look at current research in the field and use that to propose novel solutions to our pressing perception challenges. Join us, and be at the frontier of shaping the future of autonomous deliveries!

• Ideate, experiment and iterate on learning-based solutions for unique perception challenges
• Collaborate closely with team members, brainstorming and deriving creative solutions from first principles
• Leverage large datasets from heterogeneous sources (real-world, simulation, and internet-scale data) to train machine learning models
• Ship production code to train, validate and integrate models into the perception system
• Share findings and insights, fostering knowledge exchange across teams

• You must have completed the second year of your undergraduate studies. Master’s and PhD students are also eligible
• Good theoretical understanding of 3D computer vision and various camera models
• Hands‑on expertise in and in-depth understanding of one or…