- Led the end-to-end integration of Scanless Tech CI/CD pipeline, automating the build, test, and deployment phases for microservices that run on Nvidia Jetson edge devices. Developed key CDK constructs, including Infrastructure setup, ECR publishing and VPC setup, enhancing the efficiency of software delivery.- Designed an image collection buffer for the Scanless Barcode Detector (SBD) application that runs on the robotics workcell's edge compute. This buffer is equipped with flexible querying capabilities via an API, supporting a variety of filters. The buffer can be optimized based on system capabilities and for different amounts of load. This solution enabled SBD to store images for further analysis, particularly in scenarios involving failed scans.- Optimized the edge-device commissioning process by implementing a guided provisioning strategy, significantly simplifying onboarding for the Scanless Barcode Detector service onto new edge devices. These changes had a substantial impact on enhancing the onboarding experience and satisfaction of our customers.- Created a WebSocket streaming pipeline, presenting real-time process-path status for individual items in Amazon Fulfillment Centers. This visualization is based on item barcode detection output and the inventory database, providing valuable insights into operational processes.

- Designed the architecture of Smartbeats for the Glendale Police Dpt. Smartbeats is a public safety software solution that uses historical crime data to create optimal police patrol areas (beats) on an interactive open-street map. The design has two parts: First, Data Ingestion Service, where users can upload city crime data and boundary shapefiles. Second, Data Processing Service to create police patrol areas based on various input params like date-time range, priority, beat creation criteria, etc.- Developed the Smartbeats prototype Full-stack architecture utilizing Django, Flask, and AWS cloud services like EC2, S3, RDS, Aurora, and SQS.- Implemented the AWS integration workflow of Gunshot detector IoT application for the Phoenix Police Dept. In the event of a gunshot:1. The raw gunshot audio .wav files are sent to Amazon S32. The detected gunshots are triangulated to pinpoint the location 3. The UI dashboard is updated with the event information (Google map with gunshot locations, Audio player with all the gunshot raw sound playback, Form to subscribe/unsubscribe to notifications)4. Subscribers (Police Officers) are notified via text messages/email

- Designed the architecture for the migration of microservices to the docker container environment which was incorporated in every Jio cloud enterprise service. Design included: Remote volumes for config/log persistence; Environment variables to override configuration; Cronjob support; Communication with remote host/container through ssh; VIP support for container clusters.- Decreased the testing time by 70% by creating a low level data analysis tool using a shell-based data accumulator and a Java parser to analyse millions of call record files.- Built an open-source Elasticsearch wrapper library which reduced the lines of code by half for Elasticsearch DB connection creation and query/aggregation operations.- Mentored and trained team members and juniors with technical assistance, overall feedback and coaching. Docker, Linux, Shell Scripting, Elasticsearch, Netty