PROJECTS
Incident Analysis Tool, HSBC Technology India
-
Developed an Incident Analysis tool for RTB(Run-the-bank) incidents using Python and Node JS.
-
The tool is capable of isolating highly recurring incidents through text analysis, so that they can be fixed on priority to reduce the total incident count.
-
The tool also has the intelligence to preemptively decide whether an incident requires any user action to be resolved based on the learning from previous data. This is accomplished through text denoising, NLP and Machine learning.
-
Built a light-weight, user-friendly user interface using Node JS which enables users to directly upload existing data on production issues.
Computer Vision based Pothole Detection and Notification
-
Used OpenCV to implement the blob detection algorithm to detect potholes from real time videostream of the roads. Once a pothole is detected, the image of the affected area and the location is emailed to the concerned authorities, in addition to being logged in a MySQL database throughphpmyadmin to maintain a database of pothole locations.
-
The entire system was built on a RaspberryPi, Pi Camera and a GPS module. This low-cost solution can be mounted on passenger or roadmonitoring vehicles.
Progressive training of CNN
The CNN was progressively trained with increasing resolutions of CIFAR-10 images to analyse the effect of doing so on the model’s learning ability and performance- using a vanilla CNN and state-of-the-art algorithms, AllConvNet and Network-in-Network.
Kuwait National Payments Scheme, HSBC Technology India
-
Worked as the sole developer on the modernization of Payment Enquiry screens using Angular and Node JS for end-to-end development and integrated the same with the existing legacy system on IBM iSeries.
-
The entire system is loosely coupled with the legacy infrastructure, thereby enabling swift migration to a cloud platform in the future.
