Oracle

Member of Technical Staff

- Contributed to projects within Oracle Managed Cloud Services as part of the Oracle Cloud Infrastructure (OCI) team.
- Developed automation scripts using Python and Selenium for the Oracle B2C Service Platform, optimizing processes and enhancing efficiency.
- Performed testing and ongoing maintenance of an internal DNS subnet allocator tool, ensuring seamless functionality.

Google Summer of Code

Student Developer

Organization: MediaPipe (Google, LLC)

- Devised a new cross-Platform Audio event Classifier solution API utilizing the open-sourced Google YAMNet model allowing developers to integrate it as an off-the-shelf MediaPipe Solution.
- Enhanced the existing MediaPipe Object detection Solution by extending it as a Python API for greater cross- platform compatibility.
- Programmed a module for customizing Drawing styles for the MediaPipe Hand So- lution API for Python, enabling broader personalization for developers.
Project Link

Direct Offices

Freelance Developer

- Responsible for designing and building a Full Stack Web App for a US-based Real-Estate Rental Business.
- Improved response time, performance, and user experience by deploying CDN endpoints.
- Improved response time by 160% on leveraging CDN endpoints to distribute assets across geographies.
- Integrated Twilio SendGrid with Django Backend for sending important emails to customers.
- Architected the Payment Gateway with Stripe for incoming payments and sending regular payouts.
- Implemented and configured Serverless Application with the help of Azure Logic App for performing routine tasks.
Project Link

CSIR-CMERI

Undergraduate Project Assistant

- Assisted in building a smart material based soft lower-limb exoskeleton to assist soldiers.
- Built a gait phase predictor using the LSTM Model which could predict 8 human gait phases with an accuracy of ~92% and validation accuracy of ~89%.
- Built the Neural Network primarily on TensorFlow and Keras which was compiled using Adam optimizer and sparse categorical cross-entropy as the loss function.
- The Neural Network was trained and validated using the data experimentally obtained, on a batch size of 64 and 10 epochs.