Building an AI-powered coach to revolutionize sports training

Led the Messenger Client Security team to create frameworks for secure storage, and automated pipelines to proactively identify vulnerabilities in various Messenger applications. ▪ Designed and implemented a platform independent secret storage API to securely store keys, passwords etc. on messenger client devices and make it easier for developers to add new secrets. ▪ Wrote and deployed an end-to-end vulnerability management framework for 3rd party libraries used by Messenger Android and iOS clients. ▪ Wrote and a fuzzing framework for Messenger developer teams to improve fuzzing coverage for critical Messenger E2EE codebase to flag potential vulnerabilities early in the development process.

Project: “Making Information Security Intelligent with Deep Learning” ▪ Developed techniques to train deep neural networks for information security tasks like intrusion detection with minimal training data. ▪ Conducted techniques of policy-based ensemble learning and federated learning to allow collaboration among different domains. Project: “Efficient Broadcast Authentication for 4G/5G Cellular Networks” ▪ Developed broadcast authentication for connection bootstrapping in 5G networks. ▪ Built defenses for protecting the cellular paging protocol against security and privacy attacks. ▪ Prototyped the proposed schemes using USRP boards and open-source LTE implementation.Project: “Analytics at the Edge for Critical Infrastructure and Emergency Management” ▪ Evaluating the performance of various Deep Learning Image-Recognition algorithms in context of edge devices like UAVs and mobile devices. ▪ Developing techniques to improve performance of these algorithms using Hardware Acceleration in System-on-Chips.Project: “Blockchain Based Hybrid-PKI for Internet of Things” ▪ Worked on a blockchain based alternative for traditional CA-based PKI to cater to the demands of IoT and cellular devices for authentication. ▪ Implemented and benchmarked the Blockchain-PKI scheme using Ethereum and the OpenSSL cryptographic library.Project: “Fast and Scalable Authentication for Internet of Vehicles” ▪ Developed a comprehensive cryptographic hardware-acceleration framework called Hardware-Accelerated Authentication (HAA). ▪ Implemented SCRA (Structure-free and Compact Real-time Authentication) supported by hardware-acceleration, to provide delay-aware authentication in time-critical networks.Project: “Erasure coding for Drone Swarms” ▪ Worked on protocols for data-redundancy using erasure-coding on drone swarms to ensure resilience against adversarial compromise.

Developed machine learning techniques to aid in information security tasks. ▪ Developed deep learning architectures using adversarial domain adaptation to reduce the amount of training data required for training an intrusion detection model. ▪ Prototyped policy-based ensemble learning to efficiently combine outputs of decentralized ML models based on individual training metadata.

Developed machine learning techniques to aid medical professionals in identifying ▪ Trained image segmentation models for detecting cancerous lesions in mammograms with high accuracy. ▪ Designed image classification models to identify various pathologies from X-Rays with high accuracy.

Worked for the Pentesting and Digital Crimes team to secure various Visa products and services. ▪ Conducted Application and Infrastructure penetration tests for various projects. ▪ Partnered with the Digital Crimes team to track down and disrupt phishing campaigns against customers.

▪ Analysed and compared the various Software Architectures as well as Project Management models including agile and waterfall models.▪ Advised Software Development teams on their various architecture related problems and design patterns to be implemented in code.

Developed desktop and web applications for a smart electronics company with millions of users.▪ Developed modules for configuring smart lighting fixtures and sensors for residential customers.▪ Created a scheduling and alerting system for users to define routines and get notifications whenever certain user defined conditions are met.