Undergraduate Student Researcher | CVIR, IIT Kharagpur & CNeRG, IIT Kharagpur

• Pursuing my undergraduate thesis in the field of continual learning and domain adaptation, under the supervision of Prof. Abir Das.
• Developed a novel technique for class-incremental continual learning using dynamically created prompts applied on a pretrained-ViT backbone.
• Used class attributes obtained via LLM querying, to compute inter-task similarity, for better utilization of task-shared concepts. (Paper @CVPR '24).
• Developing techniques to tackle the fully test-time adaptation challenge with a continuously varying target domain (continual TTA setting).

Cloud Applications Developer Intern | Oracle

• Background: Becky is a digital assistant which allows developers to triage, get information or perform actions on pods under provisioning.
• Objective: Implementing a user recommendation feature that detects user interaction with Becky and automatically provides relevant action suggestions.
• Procedure: Developed custom components using JS and created Flask-exposed backend modules that fetch data from REST APIs and conduct analysis on them.
• Other Features: A warning message is given to the user if any of their pods exceed the average provisioning time. The user can also directly ask the bot to perform actions like enabling monitoring or SSH on pods, and notifications will be sent both via the bot interface as well as via mail.

Research Intern under Dr. Arka Bhowmik and Dr. Suman Chakraborty | IIT Kharagpur

• Developed a Chest-Xray based lung disease identification and classification model and corresponding application.
• Tried out different Deep-Learning networks and augmentation techniques to identify anomalies in the lung scans and check model efficacy.
• Developed an Android application to act as an user interface to collect patient metadata and Chest-Xray images and stored them in firebase.