Jun 2022 – Nov 2023 · 1 yr 5 mos · Redmond, Washington, United States

• Developed a highly scalable, performant and reliable service in Azure Networking team to provide
• network connectivity with higher connections per second (CPS) than the existing system by several magnitudes
• Set up and performed large scale experiments to identify bottlenecks in performance and improved the performance of the system. Added rich telemetry and monitoring to raise alert in case of issues
• Responsible for designing new features for the virtual networks with enhanced reliability and
• performance

Jul 2019 – Jun 2022 · 2 yrs 11 mos · Redwood Shores, California

• Developed and launched the application for information extraction from the documents depending on the business requirements
• Developed supervised machine learning/deep learning models (including training, fine-tuning and testing) for extraction of required information from highly diverse, form-like documents
• Improved the accuracy metrics for multiple required fields with some of them significantly higher (by around 25% to 35%) than the acceptable minimum level

May 2018 – May 2019 · 1 yr · Amherst, Massachusetts

• Guide: Prof. Tauhidur Rahman
• Spearheaded the project titled ‘Smart Refrigerator’. Developed machine learning models to detect the type of food item and whether that food item is being taken out or put inside the refrigerator
• Built an end-to-end system that collects images using multiple camera on opening the refrigerator
• Created a rich dataset of over 75000 images related to human-refrigerator interaction by collecting images from 25 human volunteers
• Detected hands in the images using Tensorflow object detection API in Python by applying transfer learning on the SSD model pre-trained on Egohands dataset

Jul 2016 – Jun 2017 · 11 mos · Noida, Uttar Pradesh, India

• Designed an internal software using Spring framework in Java by which different integrations corresponding to different teams can send data to different tools
• Developed permissions feature that can be used to determine how much change/access can be made to a corresponding team's data

May 2016 – Jul 2016 · 2 mos

• Worked on the segmentation of a given search query such that the extracted keyphrases make the most sense
• Researched on the role of semantic relation and co-occurrence frequency between the words in deciding the best segmentation of the text

Feb 2016 – May 2016 · 3 mos

• Built a model for the classification of sickled patients from normal ones from their cell samples’ images
• Worked on the detection of red blood cells in the images followed by calculation of features like roundness, solidity, etc. on the detected RBCs, which are then used for classification

May 2015 – Jul 2015 · 2 mos · United Kingdom

• Guide: Prof. Ying Liu, Cardiff University
• Designed a prototype for the extraction of product features from free text containing customer reviews
• Researched and analyzed the parameters (like term frequency, synonym check, etc.) to get the best precision and recall

May 2014 – Jul 2014 · 2 mos · Mumbai Metropolitan Region

• Developed an online portal for the campaigners to send their current progress and view others’ progress for the different activities of different marketing campaigns
• Implemented photo sharing using which the users can share images for their respective campaigns

Dec 2013 – Dec 2013 · 0 mo · Mumbai Metropolitan Region

• Developed an android app ‘World of Steel’ which was hosted on Google Play
• Involved in the web portal development for the users to view various steel products