Graduate Research Assistant
University of Notre Dame, Notre Dame, IN • May 2022 - Present
•
Developing recongnition systems agnostic to low quality
images taken from UAVs considering different body poses
and angles.
•
Leveraging query strategies in Active Learning techniques to meticulously
identify and analyze key frames, effectively extracting crucial information
about events in video content to reduce and optimize the gallery sizes for recognition
systems.
•
Analysed data-driven strategies, including OSS with
sample mean estimation (OSSM) and multi-class SVM, to
optimize face recognition gallery size.
•
Achieved substantial performance gains on 2 Datasets:
TAR increased from 86.65% to 93.66% on PaSC and Ranks
1 and 5 values on IJB-S improved from 23.53% to 67.16%
and 36.87% to 74.41%, respectively.
•
Collected 165,000 human trials for facial similarity
comparisons, extracted features from human face morph
images and modeled the features using Gaussian and Extreme
Value Theory models, followed by a detailed analysis
to identify outputs closely mirroring human responses.
Graduate Research Assistant
University of Notre Dame, Notre Dame, IN • Jan 2022 - May 2022
•
Assisted in teaching Data Structures principles to 80 students,
held weekly office hours and advised 3 undergraduate project groups
as a `Preternship Manager’.
Data Engineer
OptM Media Solutions Private Limited, Bengaluru, India • Sept 2021 - Dec 2021
•
Explored different pipelines for steam leakage detection using
various TensorFlow transfer-learning based neural network architectures.
•
Performed object detection and tracking using YOLOv5 to develop a
real-time multi object tracking system.
Deep Learning and Computer Vision Intern
OptM Media Solutions Private Limited, Bengaluru, India • Sept 2020 - Aug 2021
•
Trained, deployed and tested Tensorflow biometric identification algorithm
for automated real-time employee attendance verification.
•
Developed a secured SQL database to process and store registration information.
Machine Learning Intern
VSIX Analytical Labs Private Limited, Bengaluru, India • Jul 2020 – Aug 2020
•
Implemented a regression model to predict the Flame Ionization Detection
Response Factor with 98% accuracy.
•
Built, deployed a Django web application on the Heroku cloud for seamless
integration for client data collection/processing.
