Software Engineer

• Mar 2023 - Present

Research Assistant

• Sep 2022 - Jan 2023

Worked on development of Web Scrapping Tools Worked on creating a Google patent parsing tool using beautifulsoup python package to parse the HTML files Used Selenium WebDriver to scrape intricate websites with dynamic contents Worked on Implementation of text inconsistency software Worked on development of Web Scrapping Tools Worked on creating a Google patent parsing tool using beautifulsoup python package to parse the HTML files Used Selenium WebDriver to scrape intricate websites with dynamic contents Worked on Implementation of text inconsistency software

Machine Learning Engineer

• Jan 2021 - Jun 2022

Center for Innovation in Emerging Technology (CIET)Worked on the software development of an NLP-based SaaS tool "TraceX" built for infection control of Covid-19.Designed the algorithm for information retrieval from news articles and built the machine learning model for estimating the user's risk of COVID-19 infection after ETL processing of data, which had a precision of 0.78Worked on feature selection and hyper-parameter tuning for regularization, and trained the ML model to calculate the probability of getting Covid-19 in the user's locality, which gave an AUC score of 0.8Worked on an AI based SaaS tool for Cyberbullying Detection & Prevention on Social Media.Designed and trained the hybrid RNN-LSTM Neural Network Worked on BERT data augmentation and GPT-2 for synthetic data generation which improved the accuracy by 2.4% Led the end-to-end software development of the company's payroll management system with features such as a google based social authentication, clock-in and clock tracker etc. The Next.Js React Framework was used for enabling server-side rendering features and Node.js was used for implementation of backend logics. Show less

Director Of Administration

• Aug 2020 - Jun 2022

About Me:As a part of Feminist Pen Foundation, I aim to bring awareness to fellow people in the community about the gender imbalance that exists in the society and how it is being perpetuated due to lack of action or acknowledgement by a socially privileged majority. I strongly believe in the strength of numbers and when that unity is built among a growing generation of youths, they will grow to become non-tolerant and steadfast allies to equality. My Role:In a team of 17 employees, my role is to govern administrative staff, monitoring budgets, administer human resource requirements, disbursing fund to departments. Show less

Student Researcher

• Sep 2020 - Jun 2022

Object Recognition and Retrieval using customized ResNet Architecture: Improved the existing ResNet Model for Object Recognition by +2%, by integrating the ResNet architecture with custom-made Image Data Augmentation modules. Further, the model was made capable of Image Retrieval, using a custom-loss layer on the augmented CNN ResNet architecture Semantic Segmentation of Face Attributes using Deep-Lab based models: Modified an existing DeepLabV3+ model used for Multi-class Semantic Segmentation which, when trained and tested on a CelebAMask-HQ dataset, gave an output classification accuracy of 88.5% Model Comparison for Bank Marketing: Worked on classification model comparison for prediction of subscription to a bank's term deposit using a real bank dataset. Four classification models which are Least Square, Fisher Linear Discriminant, Logistic Regression, Perceptron were compared, among which the Least Square was the best, considering an accuracy score of 0.88 Forest Cover-Type Prediction: Implemented a machine learning model for categorical classifier for prediction of forest type. After feature scaling using MinMaxScaler, Kernel SVM and PCA-Kernel SVM model, was implemented for classification. F1 score metric was used because of data imbalance, out of which PCA kernel-SVM was the best with a score of 0.72 Background Removal using Neural Networks (MODNet) for High-quality Morphed Images created using (StyleGAN2) Face Detection & Clustering: Used Haar-Feature cascade Classifier for face detection & applied K-means clustering on the extracted 128-bin feature descriptors from each of the faces to categorize them Image Stitching using SIFT feature detection algorithm: Used SIFT algorithm to extract features from images & applied Nearest Neighbor algorithm to find matching key descriptors (RANSAC for outliers), to stitch images to create an image panorama Implementation of A-star path planning algorithm and Bug2 Obstacle Avoidance Algorithm on ROS Show less

Project Intern

• Jun 2019 - Mar 2020

Designed an algorithm for gait analysis & metrics estimation for long & high jump using an IMU sensor placed at instep of foot Developed a machine learning model for the classification of cricket (sport) shots using a wrist band IMU sensor which improved the Mean Average Precision (MAP) value by 13.8% Designed an algorithm for gait analysis & metrics estimation for long & high jump using an IMU sensor placed at instep of foot Developed a machine learning model for the classification of cricket (sport) shots using a wrist band IMU sensor which improved the Mean Average Precision (MAP) value by 13.8%

Intern

• May 2018 - Jul 2018

Indoor Navigation System for Ground Robots: Fused an MEMS sensor based Inertial Navigation System (INS) & Ultrasonic sensor using Sensor Fusion(Kalman Filter) to design an Indoor Navigation System with a true value deviation of 4.3% Indoor Navigation System for Ground Robots: Fused an MEMS sensor based Inertial Navigation System (INS) & Ultrasonic sensor using Sensor Fusion(Kalman Filter) to design an Indoor Navigation System with a true value deviation of 4.3%