(Music Expertise)
Summary:
Meta AI seeks to mobilize a panel of experts to support annotation and development of a video dataset. This dataset is composed of video data, featuring a range of skilled individuals performing a set of physical and procedural tasks around the world. We seek experts to review these video files and provide performance commentary, using deep knowledge of the skills shown to comment on each step of the activity, describe what is happening, how well the activity is being performed, and why the actions support the ultimate goals of the person doing them. This work requires expertise in the domain, effective communication, teaching experience, and commitment to the project.
Specific contributions:
With their specialized knowledge, we seek experts that can:
Provide insights into their domain of expertise via verbal commentary about a range of demonstrations shown on video and exchange with Meta AI researchers and other subject matter experts
Verbal expert commentary time stamped across videos in the dataset — offering technical perspectives on the execution of tasks by the person in the videos, tips, guidance and other suggestions
Expert qualification:
Experts must have a combination of technical training or demonstrated mastery and teaching or coaching background along with good verbal and written communications skills. A minimum of 5 years of experience in these areas of expertise should be targeted, with more preferred, along with a background in teaching, coaching, or other forms of sharing expertise. The ability to perform basic computer-based tasks is necessary. Expert should have one or more of the following qualifications:
Music:
Music degree or certification
Teaching or performance experience
Membership in National Association for Music Education
Experience teaching or creating how-to guides
Types of Instrument Experience:
-Violin
-Piano
-Guitar
Experience with multiple instruments is preferred
Hours: (Can be worked at the candidates convenience, but will be expected to be communicating updates and things with the team through workplace chat, occasional meetings, etc.)
11-15 hours per week, give or take based on surges and deliverables of the individuals video count.
Each worker will be provided a certain amount of videos to review and if they finish before those expected hours, then they would just bill for the time actually utilized to complete the coaching on those videos
This is an estimation based on the road map, and the way that it was calculated. There is flexibility in the hours adjusting based on the individuals speed as well. Please note, this is less based on how many hours needed to work but rather the deliverable of videos watched
Roadmap:
The team anticipates to have 123 hours of video total. The estimation is that this will be evenly broken out across all 8 skillset (40 positions). They would want each expert to watch approximately 15 hours of video and provide their commentary. They are thinking that doing the commentary might take 10x as long as the video itself, so they would expect each expert would work about 150 hours on the video content directly. There would then be other hours for preparing, understanding the guidelines, getting to know the project, etc.
Additional information:
We want the ability to disclose who these experts are if we want to -- for example, we might like to put their names and affiliations in a research publication
We wanted to be sure that they know -- this is 100% clear, but maybe it is good to include in the JD as well -- that all outputs will be public and open for the research community to use
We might try to make a video at some point of the experts themselves giving the expert commentary -- like the video I showed you of Chef Dominic.
We will also be asking them to record their mouse movements potentially on the videos, maybe their gaze, and of course their commentary.
