Responsibilities include:
- Researched, implemented, and trained computer vision models for sign language translation.
- Applied optimization techniques to reduce model inference time, including Input-Adaptive (IA) methods.
- Integrated alternative attention mechanisms to lower model complexity, such as SparseProb Attention and Linear Attention.
- Explored SiFormer, a state-of-the-art model for sign language recognition.
- Investigated approaches for generating sign language from spoken language input.