Worked on social fake news datasets, Extracted news context level features using W2Vec and BERT then merged them with GNN Layersto getsocial context extraction. Got a TestF1 score of 92%.

Roles and responsibilities:-

Fix software bugs by analyzing code, identifying the cause, and implementing necessary changes using SQL and HTML. -

Design and optimize database schemas and queries to ensure efficient data storage and retrieval, Perform database maintenance tasks, such as creating indexes, and knowledge graphs, optimizing queries, and ensuring data integrity using ontology.

- Collaborate with database administrators to address performance issues and maintain database and availability.

Skills: Python, NLP, Ontology, GNN, PyTorch Geometric, Knowledge graph, SQL, Word Embedding

Worked on Graph Convolutional Neural network with attention to classifying HIV using molecule-net data sets. Accuracy: 96.7%, Precision: 38.1% Recall: 43.9%, F1-Score: 40.8. Used GVAE to generate the graph-based molecules.

Roles and responsibilities:-

-Write clean, efficient, and maintainable code using Python, SQL, HTML, and CSS.

-Design, develop, and optimize database schemas and queries using Knowledge graphs and Ontology to ensure efficient data retrieval and storage.

- Perform database maintenance tasks, such as indexing, performance tuning, and ensuring data integrity.

Skills: Pytorch Geometric, Python, Graph Neural Network(GAT), Knowledge graph, GVAE, Deep Learning, Ontology

Detect product images from a shelf of goods in a supermarket. Detection widow has been trained on YOLOV4 and getsthe inference from there. For classification Ive used ResNet pre-train Network with 10 Epoch by applying TL

Extracted Various Spectral and temporal features from audio signals in order to send Multi Chanel CNN to classify themusic notes.