Apple | Cupertino, CA
Leading pre-training of LLMs catered towards edge applications.
Led multiple projects on LLM Efficiency through Multi-Token Prediction, Speculative Decoding and KV-Cache Management.
Reduced serving latency of LLMs on distributed servers by developing Sync-Point Drop.
Apple | Cupertino, CA
Developed audio-visual perception models for real-time intent understanding.
Led the development and delivery of voice trigger models for detection of wake-word on ultra low power always on devices.
Texas A&M University | College Station, TX
Developed techniques to use deep learning algorithms for unsupervised Anomaly Detection.
Loading latest research...
