Computer Scientist, with specialization in machine learning. At Google, as part of the Speech group I created the (pre-TFLite) ML inference framework used across speech and gboard related products with over 1 billion monthly active users. I also developed, from research to implementation, the latest instantiation of the "hey google" detector used across Google Assistant products, from always-on DSP chips on Android to Google Home devices. Currently in the Brain team, I lead a team under TensorFlow Lite project (which I co-founded) focused on developing techniques for efficient execution of machine-learning systems on the edge, as well as partnering with product and hardware teams to launch state-of-the-art ML across Google's products (particularly speech and text-processing related) on both commodity and specialized hardware.