Pankaj Rawat is a Master of Technology graduate in Computer Science and Engineering from IIT Madras. He gained experience as a Data Scientist Intern at Tata Communications, where he utilized Graph Neural Networks (GNNs) to predict network KPIs and build Digital Twins. His technical expertise includes deep learning frameworks (PyTorch, TensorFlow), Generative AI, and computer vision, demonstrated through projects such as Dialog Summarization and Vehicle Tracking Systems.
Experience
Intern Data Scientist
Tata Communications
Predicted geographic location of users given their IP address using Graph Neural Networks (GNN) with attention. Formulated the problem as an attributed graph node regression problem and used mean squared loss for training. Built a Digital Twin for computer networks with GNNs to predict KPIs (path delay). Utilized message passing between network nodes, links and paths along with LSTMs for accurate path level prediction. Worked with the team on research, EDA, model development, and communication with stakeholders. Used Pandas for EDA, Python, PyTorch and Tensorflow for model implementation and WandB for experimentation.
Education
Indian Institute of Technology Madras (IIT Madras)
Master of Technology
Computer Science and Engineering
THDC Institute of Hypdropower Engineering and Technology
Bachelor of Technology
Computer Science and Engineering