Bio
|| Neuroengineer || Data Scientist || Private Thoughts ||
Carlos Francisco González Rivera is a dynamic and versatile Data Scientist and Biomedical Engineer whose work bridges the cutting edges of healthcare innovation, national security, neuromorphic computing, and artificial intelligence. With a solid foundation in developing machine learning models, anomaly detection systems, and computational tools, Carlos has a proven track record in both startup environments and large-scale research organizations. His expertise spans biomedical device innovation to data-driven solutions that enhance human performance and national security infrastructure.
Career Highlights and Contributions
Carlos joined the Pacific Northwest National Laboratory (PNNL) in 2021, where he has led high-impact research initiatives across multiple disciplines, including neuromorphic computing, nuclear facility monitoring, and synthetic biology. His key projects include:
Real-time Learning with Neuromorphic Systems: Carlos collaborated with Intel, Los Alamos National Laboratory (LANL), and Drexel University to develop Spiking Neural Networks (SNNs) for real-time data processing using Intel's Loihi neuromorphic chip. His work advanced adaptive machine learning models that mimic biological processing, published in ICONS 2022.
Persistent DyNAMICS: Over three years, Carlos designed and implemented unsupervised machine learning models for anomaly detection, improving nuclear facility monitoring systems by extracting and analyzing key features using SHAP values, LIME, and timestamp rolling. His innovations in predictive monitoring were presented at the 2023 INMM & ESARDA Joint Annual Meeting.
Image Processing for Synthetic Biology: Leveraging state-of-the-art computer vision tools such as OpenCV, DINO, SAM, and Grounding DINO, Carlos developed image segmentation techniques to analyze microbial images in PNNL's Persistence Control of Engineered Functions in Complex Soil Microbiomes (PerCon SFA) Project, sponsored by the Genomic Sciences Program (GSP).
Interdisciplinary Expertise and Research Impact
Carlos's academic journey began with a Bachelor of Science in Biomedical Engineering from the Polytechnic University of Puerto Rico, followed by his pursuit of a Master of Engineering in Electrical and Computer Engineering at Rice University. His graduate capstone focused on leveraging machine learning to identify cardio-respiratory signatures that predict adverse outcomes in Sudden Infant Death Syndrome (SIDS) mouse models, achieving over 85% accuracy in outcome predictions. This work contributes to predictive healthcare, with potential applications in real-time health monitoring systems for at-risk infants. Meanwhile, at Neurotech@Rice, Carlos pioneered neurotechnology software research, developing EEG-processing workflows and teaching workshops on machine learning. His contributions here underscore his role as both an educator and a researcher, mentoring future leaders in data science and neuroengineering.
Key Technical Skills and Contributions
Carlos's technical expertise spans a wide array of fields, combining his deep knowledge of machine learning, artificial intelligence, and biomedical engineering to push the boundaries of innovation:
Machine Learning and AI: Proficient in PyTorch, TensorFlow, Scikit-learn, and custom neural networks. Carlos has implemented models for physiological signal analysis, image processing, and real-time anomaly detection.
Neuromorphic Computing and Spiking Neural Networks (SNNs): Carlos's work with Local Competitive Algorithms (LCAs) and SNNs on Intel's Loihi2 processor has revolutionized real-time adaptive learning systems, emulating biological processes and applying biomimicry inputs from Dynamic Vision Sensors (DVS) and event cameras.
Data Science and Feature Engineering: His adept use of data science techniques for feature extraction, predictive modeling, and dynamic featurization has had a significant impact on nuclear security, synthetic biology, and healthcare.
Computer Vision: Carlos excels in using OpenCV, SAM, Grunding DINO, and Grounded-SAM for tasks such as image segmentation, object detection, and biomedical image analysis.
Teaching and Leadership
Carlos is passionate about teaching and mentoring the next generation of data scientists and engineers. He taught "Python for Data Science" to high-school-level interns through PNNL's Bridging Opportunities for Leadership and Training in STEM (BOLTS) program while leading Python tutorials for Rice University's "Introduction to Neuroengineering" course. His work as a Teaching Assistant (TA) and D2K Lab sponsor has shaped the future of young engineers, providing them with the tools and techniques necessary for success in data-driven research.
Future Directions and Research Interests
Carlos's research interests continue to evolve at the intersection of artificial intelligence, neuroengineering, and computational biology. He is driven by the possibilities of predictive modeling in healthcare, where early detection of physiological abnormalities can save lives, and the potential of neuromorphic computing to revolutionize real-time adaptive systems. His work in biomimicry, using advanced sensors and neuromorphic processors, opens the door to new paradigms in human-machine interfaces, digital healthcare, and national security.
Key Tools and Technologies for Professional Portfolio:
Programming Languages: Python, C++, R, MATLAB
Machine Learning Frameworks: PyTorch, TensorFlow, Scikit-learn
Neural Networks and Algorithms:
Spiking Neural Networks (SNNs): Hands-on experience with Intel’s Loihi neuromorphic processor, implementing SNNs for real-time adaptive learning.
Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) for physiological data modeling, image recognition, and sequential data analysis.
Unsupervised Learning Algorithms: Applied clustering, density-based, and anomaly detection algorithms (e.g., ADBench with DeepSVDD, DAGMM) in high-dimensional data for nuclear facility monitoring and healthcare research.
Data Science and Predictive Modeling:
Advanced Feature Engineering: Extracted meaningful features from physiological signals and high-dimensional datasets for anomaly detection, nuclear security, and predictive healthcare models.
Statistical Analysis: Performed regression, hypothesis testing, and correlation analysis to enhance data-driven insights and model accuracy in research domains.
Model Explainability: Used SHAP values and LIME to interpret and visualize model decisions, improving the transparency and robustness of machine learning models.
Signal Processing:
Expertise in time-series analysis for physiological signal processing, including cardio-respiratory data for SIDS research and sensor data for real-time monitoring.
Applied frequency-domain and time-domain techniques such as filtering, Fast Fourier Transforms (FFT), and wavelet transforms for signal analysis and noise reduction.
Computer Vision and Image Processing:
Leveraged OpenCV, DINO (DeNoising Anchor Boxes), Grounding DINO, and Segment Anything Model (SAM) for automated image segmentation, object detection, and microbial image analysis in synthetic biology.
Applied advanced vision algorithms to process real-time data streams from Dynamic Vision Sensors (DVS) for neuromorphic computing.
Data Engineering and Preprocessing:
Developed custom pipelines for large-scale data processing using Pandas, NumPy, and SciPy to handle physiological, image, and time-series data.
Implemented Heterogeneous Euclidean Overlap Metric (HEOM) for imputing missing data in nuclear facility datasets.
Development Tools and Version Control:
Proficient in Git, Docker, and Jupyter for code versioning, containerization, and interactive development.
Experience with DVC (Data Version Control) to manage data processing workflows and experiment tracking for machine learning projects.
