- w github.com/Adarsh-gadari
- LinkedIn l adarsh-gadari (LinkedIn)
Currently, I am a PhD student in computer science at University of North Carolina, Greensboro. My research focuses on developing Artificial Intelligence (AI) algorithms. With an academic background of signal processing, communication and electronics engineering, and a professional experience of 4 years (July 2018 - July 2022) in AI across domains like engineering, medical images (Ophthalmology), medical structured data, health informatics, I would like to develop algorithms that can integrate multi modal data.
Health Informatics: Exploratory and statistical data analysis on the legacy data to identify the high risk groups. Hands-on with the on-premises data connection using Hadoop to retrieve data. Specifically, the focus was to reduce the dimensionality of the data by either creating latent variables or remove the non-contributing variables.
Return to office - tracking covid risk: Developed and deployed a solution to track the covid cases across cities and states to evaluate the risk in returning to office. Scheduled the triggers to fetch data from government websites, develop the ARIMA model and push the predictions - the number of positive cases and number of deaths for the next 14 days.
Ticket bucketing using text classification: Applied text preprocessing methods like word to vector conversion using genism library, Term frequency- Inverse document frequency, etc. followed by machine learning models like support vector machine. Saved 100K USD worth man hours by classifying the tickets automatically.
Prediction of sepsis: Identifying the onset of sepsis at least 6-hour prior. Applied recurrent neural networks in combination with Long-Short Term Memory units on the time series natured data.
Emotion detection: Identifying the emotions from a face as a use case to drive tele-health by artificial intelligence. Developed a solution using convolutional neural networks.
Introcular Lens(IOL) Power prediction: Preprocessing the huge biometry data, performed web-scraping to pull the IOL predictions of other formulas. Developed an AI regression model to predict the IOL power and corresponding post operative spherical equivalent refraction. Deployed in the LVPEI Intranet for further validation. OM 3.0, Pupil Detection: Developed an automated software to detect pupil from an Infra-Red image. Optical Character Recognition(OCR) for oculyzer images: Developed and trained an OCR to extract eye information from oculyzer images. Performed statistical analysis on the data. Extracting biometry information from A-scans: Extracted data from a set of 50,000 PDF files with different variations which include missing data, change in the format of the data, etc. Performed statistical analysis on the data and won the best poster award at comprehensive cataract conference, Kolkata 2019 |
Developed a classification model to predict the severity of a failed equipment. Hands-on experience in creating a GUI in python. Conversion of R codes to python. Prepared dashboards in Tableau.
- Infinite Family of Feature Space - Machine Learning (IFFS-ML) - Prof. Shan Suthaharan (2022-present)
- Immersive Multimedia Technology (IMT) - Prof. Soumya Jana (2016-2018)
Department of Computer Science.
Current research focuses on computational ophthalmology which includes parametrization, finding feature space, and Machine learning.
8.1/10 GPA
Communication and Signal Processing, Dept. of Electrical Engineering
Electronics and Communication Engineering
Gadari Adarsh, Chandra Bollepalli S, Ibrahim MN, Jose´-Alain S, Chhablani J, Suthaharan S, Kumar Vupparaboina K. Robust Retinal Layer Segmentation Using OCT B-Scans: A Novel Approach Based on Pix2Pix Generative Adversarial Network. InProceedings of the 14th ACM International Conference on Bioinformatics, Computational Biology, and Health Informatics 2023 Sep 3 (pp. 1-6).
Gadari Adarsh, Chandra Bollepalli S, Ibrahim MN, Jose´-Alain S, Chhablani J, Suthaharan S, Kumar Vupparaboina K. Retinal sublayer segmentation based on deep learning using optical coherence tomography B-Scans: Training multilayer masks vs boundary labels (Accepted for ARVO 2024, May 5-9)
◦ PYTHON ◦ C ◦ C++ ◦ MATLAB |
◦ Deep Learning ◦ Pretrained models ◦ Residual Networks ◦ Transformers ◦ Attention mechanism |
| ◦ Regression ◦ Classification ◦ Supervised Learning ◦ Unsupervised Learning ◦ Time series Analysis |
| ◦ Tensorflow ◦ Keras ◦ Scikit-Learn ◦ OpenCV ◦ Nltk ◦ Numpy ◦ Pandas ◦ Matplotlib |
| ◦ Tableau ◦ Git - Github - Bit bucket ◦ VsCode ◦ Docker ◦ Jenkins |
◦ Machine Learning ◦ Image Processing ◦ Signal Processing ◦ Information Theory ◦ Probability and Random Process |
◦ Third Prize in Arch MI Data Dive competition held in association with a Non-Profit organization.
◦ Consolation prize at Legato Techathon for developing automatic sign language detection using AI.
◦ Recognized by the team at Legato with for proactively proposing solutions to make tele-health AI powered.
◦ Won best poster award at Cornea conference, Kolkata -2019.
◦ GATE rank 5620 of 172714.
◦ Volunteered to Integrated Household Survey conducted by the Government of Telangana.