Research Intern at IIT Kharagpur (Dec 2015 - Feb 2016)
- Classification of features using KAZE and AKAZE descriptors for detection of microaneurysms.
- Used Active contour model to detect and segment lesions using OTSU thresholding.
- Tools used: Python and OpenCV
Software Engineer at Kalpah (Feb 2016 - April 2016)
- Worked on an automated glaucoma detection software that involved image processing on fundus images of retina for detection of optic disc and the optic cup.
- Segmented and processed contours to calculate various parameters such as DDLS, VCDR, HCDR, ISNT, and NRR.
- Worked on segmentation of vessels by applying a set of rotated Gaussian Hermite kernels over n1 matched filter orientations with a n2-degree increment.
- Developed an multi-threaded Android app for the aforementioned algorithm developed in Python and used Java's Runtime exec API for integration.
- Tools used: Python, OpenCV, Scipy, Numpy, Scikit-learn, Java, Android SDK
Lead Software Engineer at Kalpah (April 2016 - Present)
- Worked on fundus montaging software; involved working with keypoint descriptors, Homography, transformations of matrices, and blending using distance transforms.
- Significantly optimized the montaging by performing aforementioned operations on each image using concurrency.
- Worked as Platform Engineer with several clients to integrate our SDK into their applications.
- Developed an multi-threaded Android app for the aforementioned algorithm developed in Python and used Runtime exec for integration.
- Tools used: Python, Java, OpenCV, Scikit-learn, Scipy, Numpy, Android SDK
Call Utils (Jan 2015 - Present)
- Extensively worked with BroadcastReceivers, Services, and WindowsManager for blocking/whitelisting, call recording, and showing a multi-use case floating pop-up during calls
- Used the Model-View-Pattern architecture to separate the logic layer, network layer, data layer and interface layer for ease of testing.
- Uses Dagger 2 for dependency injection utilizing custom scoping of objects, RxJava, Realm and Espresso. Utilized practices held beneficial by the android community.
- An Android app that scrapes football GIFs from an endpoint running on EC2 which scrapes football media to keep users up-to-date with enticing events from matches.
- Used Dagger 2 for dependency injection, RxJava for reactive paradigm, and OKHttp.
Reddithor allows users to browse the posts from subreddit of their choice from within the tray of their Linux OS. It automatically updates every 2 minutes in the background by scraping posts from Reddit. Used PyGTK for working with UI elements, and Python's request module for scraping
StayWoke is an Android library that extensively uses ActivityLifecycleCallBacks in combination with WindowsManager to:
- Keep the screen from turning off while in specified activities.
- Wake the screen up as soon as the app gets deployed on to the phone.
- Runner up at the National Robotics Championship conducted by the department of Energy Science and Engineering, IIT Bombay.
- Call Utils was featured on XDA-developers, App Of The day, and AppGratis.
- Call utils was trending for 3 consecutive days, under the Tools category on the Play Store. Currently installed on 95,000 active devices, with a total of 110,000+ downloads.
- GIFs for Soccer was first in "Top new paid", third in "Top new free", and second in "Top paid" in the 'Sports' category on the Play Store.
- Kalpah was crowned the best startup in Andhra Pradesh by India Today — presented by the Chief Minister of Andhra Pradesh and Rajdeep Sardesai.
GITAM University (2013 - 2017)
Bachelor Of Technology in Electrical and Electronics Engineering
Functional Programming, RxJava, Java, Python, Django, OOP, C, RESTful Services, Software architecture, Software Design Patterns, Android SDK, JUnit4, TensorFlow, Machine Learning, Rust, OpenCV, Scipy, Go.