Collaborations

Predictive Maintenance for fleet management

Engineered a comprehensive and scalable predictive maintenance pipeline for fleet management, focusing on predicting failure types of assets
Tackled the challenge of unbalanced datasets by implementing oversampling and undersampling strategies, along with meticulous model tuning to achieve an improvement in predictive accuracy by 14%
Ensured efficient deployment and monitoring by automating model selection, model tuning, and feature engineering

Established a foundation for integrating computer vision models into the Nockout mobile app to deliver tailored advice for badminton players by simulating and analyzing game data
Generated realistic game datasets, including player and ball positions and shot types using a multi-agent reinforcement learning model
Developed a comprehensive analytics pipeline that visualizes game flow and provides actionable insights into player movement and shot-type effectiveness, helping players identify their strengths and areas for improvement

With Prof. Phil Scarf (Cardiff University, South Wales)

Formulated a robust statistical model to simulate emerging cricket formats like T10 and four20
Implemented web-scraping and utilized statistical methods to extrapolate data from well-known cricket formats to T10 by applying advanced data preprocessing and clustering techniques.
Gathered data from simulations to study various properties of the proposed format like outcome uncertainty, stadium impact, and batting aggression

With Prof. Rizwan Parveen (Birla Institute of Technology and Science Pilani, Goa Campus)

Developed an automated resource allocation model that simulates patient-caregiver interactions in a healthcare setup
Published the flexible model, modular user libraries and observations in the IEEE SMC 23 conference in a paper titled
"Building Extensible Model for Healthcare Resource Allocation using Model-driven Approach"

With Prof. Hemant Rathore (Birla Institute of Technology and Science Pilani, Goa Campus)

Analyzed and reviewed multiple papers on GAN-based Malware Detectors.
Examined numerous malware detection models and explored various attack and defense literature.

Developed an Android application that enables users to shop safely during the COVID lockdown. Users can place and track orders within the safety of their homes and collect items from the store at the allotted timeslot optimized for avoiding crowds.
Implemented the app using Java with Android Studio, certified the APK, and launched on the Playstore.

This is a one-stop app for Bitsians to fulfill their campus needs. The app has features for ordering food, viewing exam schedules, purchasing campus merchandise, organizing carpools, and a lot more.
This opportunity was my first dive into software development. It was also the first time I had to collaborate with others. Along with learning the fundamentals of Android Studio (Java) for mobile app development, I learned how to make wireframes on Figma, version control with Git, and task management with notion.

Page updated

Google Sites

Report abuse