In Advanced Database Systems Implementation course, I created this interactive 3-D visualizer that lets the developer analyse how two approximate-nearest-neighbour (ANN) algorithms perform in the same vector space.
When the developer enters a word and the web application does the following:

1. Projects its GloVe embedding (100D) into a PCA-reduced space.

2. Runs the query through classic HNSW (layered, greedy top-down) and through ACORN-1 (radius-refined walk on the base layer).

3. Shows every hop, layer switch, and radius check in parallel, so the developer can spot where paths diverge or converge.

4. Logs traversal time, hop count, and cosine similarity for the live query and ten random words, then rolls up the averages in a compact result dashboard.

5. The result is a visual analyse of accuracy and performance for anyone modifying ANN graphs for specific application in vector database.

Technologies Used:  D3.js, Python (Flask), SBERT (Sentence Transformer), PCA (Principle Component Analysis), REST API, Flask 

LinguaLink is a web application designed to facilitate effective communication between users who speak different languages. Traditionally, users who speak different languages have to constantly search for translations and copy-paste text to communicate with each other. This process can be cumbersome and time-consuming. LinguaLink addresses this issue by allowing users to directly translate text into any language they desire, streamlining the communication process.

The application leverages the Gemini API to provide translations that not only convert the text but also help users understand metaphors, idioms, and figures of speech in the target language. This feature ensures that the nuances and subtleties of the original language are preserved, making the communication more accurate and meaningful.

With LinguaLink, users can seamlessly interact with others without the need for external translation tools, enhancing their ability to connect and understand each other across language barriers.

Technologies Used:  ReactJS, HTML and CSS, Gemini, Firebase, Git, Figma, Miro,

I did research in Georgia Tech's Sonification R&D Lab under Prof. Bruce Walker to explore the integration of 3D audio into a virtual reality (VR) project called Strangeland VR. 

The project utilized Git for version control and code management. To enhance accessibility and spatial audio navigation within the VR environment, we implemented binaural audio using the SWAN 3D audio system. 

Additionally, we created Navmesh on walkable spaces for pedestrian NPCs (non-player characters) to define a customized testing path for the SWAN 3D audio system.

Technologies Used: Unity, Git, SWAN 3d Audio