Nikunj Aggarwal

Staff Software Engineer at Citizen

Nikunj leads the machine learning team at Citizen mainly working on speech processing from radios and reducing hate speech on the chat platform. Prior to this, Nikunj worked on Uber’s open sourced distributed time series database (M3), was one of the core contributors to the ML platform Michelangelo and also helped rebuild the realtime dispatching engine. Prior to Uber, Nikunj worked at Amazon on the Kindle ecosystem. Nikunj holds a Bachelor’s degree in Computer Science from IIT Madras.

WATCH LIVE: December 9 @ 3:25PM – 3:55PM ET

Opening up the 911 system using a Convolutional Neural Network

Citizen scans thousands of public first responder radio frequencies 24 hours a day in major cities across the US to provide real-time safety alerts about incidents like fires, car collisions, and missing persons. To process thousands of hours of daily radio calls in a scalable manner, we built an inhouse deep learning-based solution for detecting safety incidents. The talk will focus on lessons learnt around building such a system starting from publicly available components, and how we ended up creating an end to end safety alert system from scratch using a novel application of Convolutional Neural Networks.