MotionSavvy is bringing 3D sensing technology to the deaf community to provide real time sign language interpretting in a small and effecient form factor. It's product name UNI is designed to run on standard off the shelf hardware such as windows surface pro 4 tablets.
MotionSavvy has been recognized in numerous publications for it's innovative work that impacts roughly 360M+ deaf individuals. Being deaf in today's world is like moving to a country that speaks a different language and never being able to learn the language spoken.
To figure out what someone is signing, you must first figure out where the signs are. This can be a bit tricky, especially since moving from one sign to another involves moving the hands from one ending position to another starting position in a way that is pretty meaningless but much more substantial than equivelent features of speech or cursive handwritting, for example. Segmentation problems are often a bit tricky, though in our case the fact that we are dealing with time series data rather than, for example, images, simplifies things somewhat. We're looking for someone to take the lead on this challenging set of problems.
- Develop and improve techniques for segmentation in the context of signs and fingerspelling
- Develop and improve techniques for distinguishing signs and non-sign movements
- Help the machine learning team with other tasks as required
- Substantial experience with segmentation or sequence labelling tasks, especially in the context of time series data
- Proficient with Python
- Familiar with Git or another modern, distributed version control system
- NoSQL database experience is a plus
- Significant research experience is a plus
- ASL proficiency is a plus
- $16/hour base pay
- 10-15 hours per week