Kanpur Nagar, Uttar Pradesh, India
Wanna work on intersection of tech, psychology and brands. Wanna talk on anything ranging from economics, ml, tech, philosophy? Hit me up in DM. Interested in Software Dev/Machine Learning/Quant Roles. Currently exploring Blockchain, Crypto, Concurrency, Distributed Systems, Machine Learning and Quantitative Finance.
Walmart Plus Division. Walmart Plus Customer Acquisition Team which helps with Walmart Plus membership
Worked end-to-end on development and testing of a Java based Spark application to perform deletes over a range of Databases. • Setup the whole system to monitor and capture the metrics from Spark Jobs(on GCP Dataproc) to Graphana Dashboard. • Worked on a Spring-Boot based design of the same application as well. • Tech Stack - Java, SpringBoot, GCP, Apache Spark, Apache Kafka, Apache Airflow, Dataproc
1. Designed, developed and integrated microservices on GCP. 2. Created API Gateways to secure the API endpoints and also installed Load Balancers to configure custom domains. 3. Worked on authentication for the platform. 4. Wrote Async code in python to speed up I/O bound operations in microservices. 5. Created and deployed microservices on GCP Cloud Run using Docker containers. 6. Also participated in ETH Denver Hackathon. Worked on backend for developing a clone of OpenSea, with some added features for personalization. 7. Scraped data for the platform using selenium.
1. Worked on the Data Engineering side of the ‘Conversational AI’ team at Walmart. 2. Worked on setting up a data pipeline to collect data from Kafka Topics and save them in a Cassandra Table after doing the required processing. 3. Worked on the logic to select the table specifications in Cassandra, according to the requirements of the service making the queries on the data. 4. Explored internals of Kafka and implemented the concept of Manual Offset commits in the prototype application. 5. Implemented the prototype application using Java clients for Kafka and Cassandra.
1. Worked on the optimization of the image pre-processing part of the OCR pipeline 2. Used the concept of Super-Resolution of images to improve the readability of low-resolution document images using Deep Learning based Generative Model - SRGAN 3. Developed a command-line utility to train the SR-GAN model for 2x, 4x.. resolutions and to super-resolve test images for different resolutions. 4. Understood the concept of transfer-learning for extracting features from an image, and hence fi ne-tuned ResNet18. 5. Worked on the pre-OCR document classi fication methods to classify the documents. Implemented and ne-tuned ResNet18 and VGG16 for the task of document classi cation and achieved an accuracy of 92% on RVL-CDIP test dataset using an ensemble model of ResNet18. 6. Developed a command-line utility to train the model to different regions of page and then make a combined prediction for the task of document image classifi cation.