Sep 2022 – May 2024 · 1 yr 8 mos · Atlanta, Georgia, United States

• Graduate Teaching Assistant for the course Deep Learning (CS 7643) under Prof. Zsolt Kira.
• Responsibilities include facilitating students' understanding of the subject matter, holding office hours, and grading.

Sep 2022 – May 2024 · 1 yr 8 mos · Atlanta, Georgia, United States

• Advised by Prof. James Hays
• Design and explore methods for 3D Object Composition and 3D Semantic Segmentation using Neural Radiance Fields (NeRFs).
• Improving multi-scene generalization of Neural Fields for 3D semantic understanding.

Mar 2021 – Aug 2022 · 1 yr 5 mos · On-site

• Designed a Dual Stage GAN architecture for high-res image-to-image translation on smartphones. Employed attention innovatively, reducing parameters by 60x and FLOPs by 9x compared to prevailing methods. Leveraged synthetic data for improved model generalization. Published at IEEE/CVF WACV 2023.
• Collaborated with a student team for two successive terms through Samsung PRISM program. Optimized a SOTA Semantic Image Editing generative model by feature sharing, enabling 13% faster inference with minimal loss of quality. Published at IEEE ISACC 2023.
• Onboarded and guided new team members technically.
• Promoted algorithmic thinking and problem solving as an active mentor of SW Competency Taskforce.

Jun 2019 – Feb 2021 · 1 yr 8 mos · On-site

• Deployed a low-latency Interactive Segmentation model on smartphones with TFLite. Improved accuracy by 2% using a multi-faceted objective that combined shape and pixel based losses.
• Built an innovative Android app to synthesize photorealistic images based on user inputs using GANs.

May 2018 – Jul 2018 · 2 mos · Dhanbad-Ranchi Area · Remote

• Analyzed several multimodal Vision Language model architectures (CNN-RNN) for Image Captioning.
• Achieved a top BLEU-1 score of 60.7 on Flickr 8K dataset with LSTM sans attention.