I lead Creator and Ads Monetization teams at Meta, enhancing creator experiences and incentivizing creators to deliver great content
Before, I managed tech readiness for Amazon events globally (Prime Day, Black Friday etc.), optimizing 200+ services for customer experience. Prior to Amazon, I built Content and Publishing Tools at Audible. Additionally, I collaborated with astronomers at NASA on the monumental LSST software for the largest telescopic camera on earth at the Rubin Observatory, Chile.
Arizona-NOIRLab Temporal Analysis and Response to Events System (ANTARES), a software designed to manage extensive streams of astronomical time-domain alerts. Given the proliferation of large-format CCDs on wide-field telescopes, nightly time-domain surveys now reveal a substantial number of new events, surpassing astronomers individual assessment capacities. ANTARES serves as an event broker, processing alerts by adding catalog references, applying filters, and identifying customizable event subsets. The article discusses the systems data model, architecture, annotation, filter implementation, outputs, tracking, performance, and user interface in detail.
The Large Synoptic Survey Telescope (LSST) will detect an unprecedented volume of transient events, necessitating a new follow-up approach. Automated alert-brokers like ANTARES will manage these alerts. A machine learning pipeline is developed for multiband optical photometry-based classification of variable and transient sources. The pipeline includes variable vs. transient categorization, multiclass typing, and purity-driven subtyping. Performance is validated on actual observations, showing competitive results. The pipeline is being adapted for real-time use on live alert streams from time-domain surveys, signifying a significant step towards advanced event management.
The frequent emergence of high-dimensional data poses a challenge in underwater wireless sensor networks, as classical clustering techniques struggle with such complex networks of arbitrary shapes. This is due to their parameter dependency and inability to handle arbitrary shapes effectively. Traditional clustering methods fail to address the high-dimensional nature of networks with arbitrary shapes and enhance clustering performance. In response, researchers have previously explored fuzzy logic-based clustering approaches. This study evaluates these techniques to identify trends. The paper introduces a fuzzy-based algorithm for clustering ensemble, categorizing sensor nodes into multiple clusters based on major and minor criteria. The algorithm effectiveness is measured through clustering precision and its capability to differentiate between network nodes. Existing clustering algorithms for data aggregation are reviewed and compared with the proposed approach.
The study highlights energy consumption as a significant challenge in sensor network design. Underwater sensor networking facilitates applications like environment monitoring and exploration. Sensor nodes collect and transmit data to a central point. The report reviews recent data collection algorithms and clustering schemes in underwater sensor networks, aiming to fulfill ongoing research requirements. Energy-efficient data aggregation for long-term network monitoring is a key focus. The study presents algorithms that address deployment, localization, and varying conditions in underwater sensor networks.
Read article
Global Citation Distribution Map
Press
My work on NFTs garnered attention from media publications globally including
