Developed and ensured efficient end-to-end ML/AI & applications pipeline orchestrations geared for final production release.
• Used scalable cloud tools to collect, preprocess, inspect and improve data quality from various sources for model training.
• Used Vertex AI for continuous ML model training and deployment, integrating with Google Cloud for scalability.
• Developed AI-powered chatbots using both Google’s PaLM API and CloudRun and with Amazon Bedrock, leveraging advanced frameworks like LangChain, Faiss, and Streamlit for diverse applications including PDF-based FAQs.
• Monitored applications and ML models to preemptively identify and resolve issues.
• Resolved CI/CD pipeline issues to ensure timely product delivery.
• Collaborated with principal scientists to conduct communication overhead analysis for adaptive resource management in deep learning applications.
• Utilized NERSC’s Perlmutter supercomputer for in-depth communication tracing and analysis of coupled simulation and deep learning workflows.
• Developed performance models to optimize memory usage, data movement, and communication costs.
Machine Learning Researcher
Oak Ridge National Laboratory
• Collaborated with principal scientists to design and develop an AI framework for detecting fabrication and masquerade attacks in the Controller Area Network (CAN) bus, enhancing vehicle cybersecurity.
• Demonstrated expertise in in-vehicle security through advanced techniques such as graph embeddings and decoding CAN log files using DBC files.
• Utilized Python and libraries including NetworkX, Pandas, Matplotlib, Seaborn, and Scikit-learn to conduct comprehensive data modeling and cybersecurity analysis.
Computer Science Research Associate
University of Texas at El Paso
• Developed condition monitoring AI algorithms for cyber-physical systems, enhancing anomaly detection and improving predictive maintenance.
• Engineered federated learning models for distributed network anomaly detection, optimizing data privacy and security across decentralized systems.
• Designed adversarial sampling techniques to test and enhance the fairness and robustness of deep neural networks, ensuring equitable and unbiased AI-driven decision-making.
• Implemented blockchain-based AI frameworks for secure identity management and data flow in critical infrastructure, contributing to the integrity of smart and connected systems.
• Leveraged Graph Neural Networks (GNNs) to enhance network anomaly detection, significantly increasing the accuracy of identifying complex threats in large-scale data environments.