Aug 2023 – Dec 2023 · 4 mos · Gainesville, Florida, United States · On-site

• Teaching Assistant, supporting students in their academic journey through the CAP 5705 - Computer Graphics course offered by Prof. Alireza Entezari
• Assist with logistics and support for smooth delivery of the course

Apr 2023 – Dec 2023 · 8 mos · Gainesville, Florida, United States · Hybrid

• Execute prior-art searches on technologies developed by UF researchers. These searches help in making informed decisions to market technologies for licensing.
• Create marketing content that explains the features and functions of various technologies, with the aim of licensing them to potential investors and industries.

Feb 2022 – Jul 2022 · 5 mos · Bangalore Urban, Karnataka, India

• As part of the Beech team, worked on upgrading the Back2Back User Agent of Cisco Webex to meet IL5 security standards.
• Upgraded scripts from Python 2 to Python 3, migrated services from Red hat OS to CentOS, updated underlying libraries to the latest versions and applied necessary security patches to align with latest requirements and enhance security.
• Designed and developed a new optimized Continuous Integration (CI) pipeline on Jenkins using declarative syntax.
• Optimized the pipeline by improving resource allocation, parallelizing tasks, and pre-compiling specific components to reduce overall build time.
• Successfully implemented the 1st stage of the pipeline, resulting in a significant time reduction from 1.15 hours to just 18 minutes.

May 2021 – Jul 2021 · 2 mos · Bengaluru, Karnataka, India

• Part of a 2-member team that developed a novel approach for internet traffic classification.
• Proposed a multi-model approach for efficiently classifying packets of a given traffic flow into either of the 4 categories - Browsing, Video streaming, File transfer, VoIP.
• Trained the model on a custom created dataset of 780 flows captured using the Wireshark tool and designed a LeNet-5 style CNN architecture.
• The proposed solution used only 15 seconds of data outperforming the then popular approaches which used around 60 seconds of data. Co-authored a research paper discussing our methodology and results.

Apr 2021 – Jul 2022 · 1 yr 3 mos · Bengaluru, Karnataka, India

• Part of a 4-member team that focused on developing a context-aware slot tagging model for Natural Language Understanding.
• Gained extensive knowledge of various NLP tasks, including different encodings, LSTM networks, and explored state-of-the-art models like transformers, BERT and its variations, and the BiDAF model.
• Identified the challenge of poor performance of models on unseen data and proposed incorporating contextual information into the model.
• Designed a modified version of the BiDAF model to incorporate labeled sentences along with related contextual information as inputs.
• Trained the model using the MIT movie corpus dataset and obtained corresponding contextual data through Wikipedia crawling.
• Achieved a notable improvement over the existing model, with correct outputs on previously failed examples.
• Recognized with a certificate of excellence and an internship offer at SRIB.