Parampreet Singh Email & Phone Number

United States

United States

United States

• Boosted predictive accuracy of vacancy formation energies by 15% through the application of a modified Crystal-Graph CNN, leveraging residual blocks, custom convolutional layers and transfer learning, resulting in an R2 score of over 0.89.• Developed a generative domain translation model using Graph Convolutional Neural Networks and Generative Adversarial Networks to predict DFT-relaxed crystal structure properties, reducing computational costs, and enhancing high-throughput virtual screening.• Implemented and optimized an end-to-end data pipeline using pymatgen, ASE, and PyTorch for data loading, feature extraction, and data collation, ensuring efficient handling of variable-sized graph data and high-quality input for model training and validation.• Built a multi-source AI knowledge agent leveraging Retrieval-Augmented Generation (RAG), integrating LangChain, Chroma, and Hugging Face Embeddings to intelligently retrieve and interact with documents from Wikipedia, Arxiv, and a vector store database.

United States

• Reduced unexpected IT asset downtime by 25% by building a predictive maintenance model using Azure Machine Learning.• Enabled efficient preprocessing of large-scale historical IT asset data, leveraging an Azure-based ETL pipeline with SQL.• Increased operational efficiency by designing a Power BI dashboard visualizing maintenance insights and schedules.• Accelerated ticket resolution time by 20% developing a GenAI-powered IT assistant chatbot using Azure OpenAI with RAG pattern for intelligent data retrieval.

Gurugram, Haryana, India

• Optimized lead prioritization by engineering a daily batch prediction pipeline using Azure Data Factory and Azure ML, processing over 10,000 sales opportunities daily.• Implemented a bi-weekly model retraining workflow in Azure ML, improving prediction accuracy by 15% month-over-month and adapting the model to evolving sales patterns.• Integrated prediction results into Azure SQL Database for CRM accessibility, increasing sales engagement efficiency by 20%.• Designed an end-to-end data pipeline with Azure Data Factory, enabling seamless data ingestion, transformation, and daily batch predictions with 99% uptime.

Chennai Area, India

1. Developed a predictive mathematical model using MS Excel to forecast the distribution of frictional power loss at the component level, i.e., bearings, gears, and seals, in a live rear axle.2. Devised and built a test rig while coordinating with different suppliers, managing timelines, and ensuring smooth cross functionality of all the stakeholders involved in the process.3. Performed a sensitivity analysis on experimental data using MATLAB to identify critical design parameters in components that most significantly affect power loss, thereby increasing efficiency by 3%.4. Published a research paper in SAE International on - Analysis and Evaluation of the Power loss Distribution Experimental data

Master Of Science - Ms, Computer Science, 3.89 / 4.0

Courses - 1. Introduction to Artificial Intelligence 2. Statistical Methods 3. Advance Machine Learning 4. Deep Learning in Computer.

Bachelor Of Technology - Btech, Mechanical Engineering, 8.2/10

Activities and Societies: Society of Automotive Engineers | American Society of Mechanical Engineers | Punjabi Editorial Board | PECFest.

12Th Standard, Non-Medical, 93.2%

10Th Standard, Science And Maths, 90.4%