# Wisdom O. Ikezogwo - Ph.D. Candidate in Computer Science & Engineering | AI Research Scientist Source: https://hello.cv/kosisoikezogwoprofile Seattle ## Links - GitHub | https://wisdomikezogwo.github.io ## About Highly accomplished Ph.D. Candidate in Computer Science & Engineering with a strong foundation in Artificial Intelligence, Machine Learning, and Electrical Engineering. Specializes in Multimodal AI, Generative Models, and Data Curation, with a proven track record of leading impactful research in medical imaging, egocentric data, and financial analytics. Leverages deep expertise in neural networks, large language models, and advanced data processing to develop innovative solutions and drive significant advancements in AI research and applications. Published extensively in top-tier conferences and journals, demonstrating a commitment to advancing the field. ## Work ### Research Assistant – Graphics and Imaging Laboratory (GRAIL) | University of Washington https://grail.cs.washington.edu/ Driving cutting-edge research in multimodal AI, focusing on medical imaging and advanced generative models. - Led research in Multimodal Large Language Models (LLMs) for medical imaging, developing foundational medical multimodal datasets (Quilt-1M, MedNarratives) with over 1 million image-text pairs. - Engineered state-of-the-art multimodal AI models (QuiltNet, Quilt-LLaVA) for comprehensive medical image analysis, achieving superior performance. - Pioneered a multi-agent AI framework, PathFinder, for clinical diagnosis, outperforming human experts in diagnostic accuracy. - Developed robust benchmarks, MedBlink, to evaluate the performance of multimodal medical AI models. - Leading initiatives to enhance the 'physics' in image and video generative models, focusing on Newtonian physics principles to generate large-scale video scene graph datasets and pipelines. ### Ph.D. Machine Learning Research Intern | Apple https://www.apple.com/ Conducted advanced research on efficient multimodal representations for egocentric data. - Led research to develop efficient multimodal representations for egocentric data, integrating video, text, audio, IMU, and hand data. - Developed 'Perceive-Predict' framework, leveraging predictive coding between co-occurring modalities to reconstruct missing data. - Applied research to significantly reduce the capture cost of expensive modalities like video, demonstrating practical impact on data collection efficiency. ### Ph.D. Quantitative Health Sciences Intern | Mayo Clinic https://www.mayoclinic.org/ Contributed to the development of foundational AI models for histopathology. - Led research efforts to develop a foundational model for histopathology, training on millions of gigapixel-sized histology images. - Scaled-up computational resources on the Argonne National Lab computing cluster to process and analyze massive datasets. - Clinically evaluated developed models, ensuring their efficacy and relevance in real-world medical applications. ### ML Engineer | Okra, Inc. https:// Developed and deployed machine learning models for financial data analysis. - Developed robust models for extracting critical customer financial information from unstructured banking data. - Identified and extracted key customer earning and spending data to feed into downstream lending systems. - Implemented predictive models for income, spending patterns, and financial reconciliations, enhancing lending decision accuracy. ### Data Scientist / ML Engineer | Demz Analytics Limited https:// Designed and implemented production-ready recommendation systems. - Developed production recommendation systems leveraging advanced techniques such as attention mechanisms and epsilon-greedy bandit strategies. - Improved user engagement and system efficiency through data-driven insights and model optimization. ### UG. Research Assistant – Biosignal Processing, Inst. & Control Lab | Obafemi Awolowo University https://www.oauife.edu.ng/ Conducted research on biosignal processing and neural network applications. - Integrated disparate multivariate time series data, characterized spectral components for dynamic dimensionality reduction. - Trained neural networks for robust classification of brain EEG signals, enhancing diagnostic capabilities. - Performed detailed characterization of EEG spectral components to understand brain activity patterns. ## Education ### University of Washington | Computer Science and Engineering 3.97/4.00 | https://www.cs.washington.edu/ - Advanced Machine Learning - Deep Learning - Computer Vision - Natural Language Processing - Data Programming - Introduction to Artificial Intelligence ### Obafemi Awolowo University | Electronic & Electrical Engineering 4.73/5.00 (Class Rank 2/120) | https://www.oauife.edu.ng/ - Digital Signal Processing - Control Systems - Electromagnetics - Circuit Theory - Microprocessor Systems ## Awards ### Population Health Initiative AI Pilot Research Grant Award University of Washington Awarded a grant of $100,000 for innovative AI research in population health. ### Microsoft's Accelerate Foundation Models Research Grant Microsoft Received a $20,000 grant to advance research in foundational AI models. ### IBRO-Simons Computational Neuroscience Summer School Travel Grant International Brain Research Organization (IBRO) & Simons Foundation Awarded a travel grant to attend the Computational Neuroscience Summer School in Cape Town. ### Prof. Kehinde Prize for the Best Graduating Student in the Control Option Obafemi Awolowo University Recognized as the top-performing student in the Control Systems specialization. ### Oyebolu Prize for Best Male Graduating Student Obafemi Awolowo University Awarded for outstanding academic achievement as the best male graduating student. ### Federal Government Scholarship Award, Nigeria Federal Government of Nigeria Recipient of a national scholarship with a cumulative value of $1500 for academic excellence. ### Total/NNPC National Merit Scholarship Total/NNPC Recipient of a national merit scholarship with a cumulative value of $1500. ### Etisalat Nigeria Merit Scholarship Etisalat Nigeria Awarded a merit scholarship valued at $250. ## Publications ### Quilt-LLaVA: Visual Instruction Tuning by Extracting Localized Narratives from Open-Source Histopathology Videos IEEE/CVF Conference on Computer Vision and Pattern Recognition | https://openaccess.thecvf.com/content/CVPR2024/html/Ikezogwo_Quilt-LLaVA_Visual_Instruction_Tuning_by_Extracting_Localized_Narratives_from_Open-Source_CVPR_2024_paper.html Introduced a novel approach for visual instruction tuning using localized narratives from histopathology videos, enhancing multimodal AI understanding in medical contexts. ### Quilt-1M: One Million Image-Text Pairs for Histopathology NeurIPS | https://neurips.cc/virtual/2023/oral/68032 Presented a large-scale dataset of one million image-text pairs specifically curated for histopathology, significantly advancing research in medical multimodal learning. (ORAL presentation) ### Multi-modal Masked Autoencoders Learn Compositional Histopathological Representations Machine Learning for Health (ML4H) | https:// Explored the effectiveness of multi-modal masked autoencoders in learning compositional representations for histopathological images, contributing to robust medical image analysis. ### Risk Stratification of Solitary Fibrous Tumor Using Whole Slide Image Analysis LABORATORY INVESTIGATION, ELSEVIER SCIENCE INC | https://www.nature.com/articles/s41379-023-01306-0 Developed a method for risk stratification of solitary fibrous tumors using whole slide image analysis, providing a valuable tool for pathological assessment. ### Recent Advances, Applications and Open Challenges in Machine Learning for Health: Reflections from Research Roundtables ML4H Symposium | https:// Contributed to a comprehensive reflection on recent advancements, applications, and challenges in machine learning for health, summarizing key discussions from research roundtables. ### Supervised domain generalization for integration of disparate scalp EEG datasets for automatic epileptic seizure detection Computers in Biology and Medicine | https://doi.org/10.1016/j.compbiomed.2020.103757 Developed a supervised domain generalization approach to integrate diverse scalp EEG datasets, improving automated epileptic seizure detection across varied data sources. ### Empirical Characterization of the Temporal Dynamics of EEG Spectral Components International Journal of Online and Biomedical Engineering (IJOE) | https://www.researchgate.net/publication/340989047_Empirical_Characterization_of_the_Temporal_Dynamics_of_EEG_Spectral_Components Conducted an empirical study to characterize the temporal dynamics of EEG spectral components, providing insights into brain signal patterns. ## Languages - English | Native ## Skills ### Machine Learning Expert - Deep Learning - Generative Models - Multimodal AI - Computer Vision - Natural Language Processing - Neural Networks - Recommendation Systems - Predictive Modeling - Domain Generalization ### AI/ML Frameworks & Tools Proficient - PyTorch - TensorFlow - Scikit-learn - Hugging Face Transformers - OpenCV ### Data Science & Analytics Expert - Data Curation - Data Analysis - Big Data - Statistical Modeling - Data Visualization - Feature Engineering ### Programming Languages Expert - Python - MATLAB - C++ ### Medical AI & Biosignals Expert - Medical Imaging - Histopathology - EEG Signal Processing - Clinical Diagnosis - Biosignal Processing ### Research & Development Expert - Scientific Writing - Experiment Design - Algorithm Development - Prototyping - Academic Publishing ### Cloud & Distributed Computing Competent - Argonne National Lab Computing Cluster (ANL) - Cloud Platforms (conceptual understanding) ## Interests ### Research - Generative Modeling - Multimodal Representation Learning - Data Curation - Medical AI - Computer Vision - Natural Language Processing ## Projects ### PathFinder: A Multi-Modal Multi-Agent Framework for Diagnostic Decision-Making in Histopathology https:// Developing an innovative multi-modal, multi-agent AI framework to enhance diagnostic decision-making in histopathology by integrating various data types and AI agents. ### MedicalNarratives: Connecting Medical Vision and Language with Procedural and Localized Narratives across all medical imaging domains https:// A project focused on bridging medical vision and language understanding through procedural and localized narratives to create a unified framework for medical imaging analysis. ### MedBlink: Probing the Fundamental Medical Imaging Knowledge of Multimodal Language Models https:// Investigating and evaluating the core medical imaging knowledge embedded within multimodal language models to identify strengths and limitations. ### Percieve-Predict: Modality and Time-Aware Egocentric Efficient Multi-Modal Representations https:// Developing efficient multi-modal representations for egocentric data (video, text, audio, IMU, hands) by leveraging predictive coding to reconstruct missing modalities. ### VPhysics: Temporally consistent Physics in Video (multiframe) Generation via Alignment https:// Research on generating temporally consistent video frames that adhere to Newtonian physics principles, leveraging alignment techniques. ### Synthetic Video Scene Graph Generation https:// A project focused on generating synthetic video scene graphs, crucial for training and evaluating advanced computer vision models. ## Source Read this profile on Hello.cv: https://hello.cv/kosisoikezogwoprofile Create your free profile at https://hello.cv