We are hiring data annotators to help us review and label our video data. Our company builds machine learning models that predict, in real-time, pedestrian awareness and crossing intent in driving videos. The training of these models relies on annotated information that we extract from clips that show pedestrians in road environments. This annotated information includes, for example, whether an object is a pedestrian, whether it is daytime or nighttime, whether the pedestrian wants to cross the road in front of the vehicle, and what a reasonable driving action would be in a given scenario. We rely on annotators to make sure that we extract that information correctly.

In this annotation role, you will be asked to scrutinize our data (sometimes in the form of images, sometimes short video clips), ensure that it has been accurately and thoroughly categorized, and, for some tasks, assess how a reasonable human driver would drive in relation to a specific pedestrian and/or the overall traffic situation. Some of these annotation tasks are more complex than others and you will be trained to annotate well.

Some of the annotations will be performed using a web-based interface such as Google Sheets, others will be performed using a desktop application that we’ve built in-house. As such you will need to have reliable access to the internet, a quality display to ensure that you can see sufficient detail in the images and clips, and a recent-model desktop computer (Windows, Mac, or Linux) on which you can install our annotation software. Higher-spec laptops are also acceptable.

You will have the flexibility to schedule your work hours during the day/week, but we expect you to commit at least 10 hours per week to these annotation tasks.

Strong Preferred Skills and Background:

• High school diploma
• 5+ years of mostly urban/city driving experience
• Basic technical and computer skills
• Previous annotation experience a plus
• Passion for data, efficiency, and accuracy.
• Ability to deliver accurate annotations at scale
• At least 10 hours per week
• Quality display to ensure you can view sufficient detail in the images and clips

Company Overview:

Perceptive Automata is a machine learning and computer vision company started by Harvard and MIT scientists that is developing a novel approach to help make autonomous vehicles safer for others on (or near) the road. We are supported by top Silicon Valley investors and global car makers and are growing rapidly in response to increasing demand for this technology.