AI Researcher
Computer Vision · Medical AI · Representation Learning · Explainable AI · Foundation Models

I am an AI researcher working at the intersection of Computer Vision, Medical AI, Explainable AI, and Foundation Models. My research focuses on representation learning, independent implementation of scientific methods, and the design of rigorous, reproducible experimental pipelines.

I have practical experience with large-scale foundation-model representations. In a collaborative satellite-imagery project, I was responsible for working with Galileo and AlphaEarth embeddings for crop classification and agricultural mapping. In parallel, my work includes medical image analysis, biomedical graph learning, and the development of explainable decision systems.

I am currently advancing research in two complementary directions:

1. Explainable AI for complex structured data, focused on construction project health assessment
2. Explainable methods for MRI brain tumor segmentation

My long-term research goal is to contribute as a specialist in AI for Cancer, with emphasis on medical image understanding, trustworthy models, and clinically relevant decision support, and to pursue postdoctoral research in this domain.

• Computer Vision — Semantic segmentation, image reconstruction, super-resolution, morphological image analysis
• Medical AI — Medical image analysis, MRI-based analysis, biomedical graph learning
• Representation Learning — Embedding design, foundation-model representations, transfer evaluation
• Explainable AI (XAI) — Interpretable models, transparent inference, uncertainty-aware reasoning
• Foundation Models — Large-scale pretrained representations, including Galileo and AlphaEarth
• Graph & Geometric Learning — Graph neural networks, graph kernels, optimal transport

Medical AI & Biomedical Learning

• Sliced Wasserstein Graph Kernel for Brain Network Diagnosis
  Independent reproduction of a MICCAI 2023 graph-kernel method using spectral graph representations, optimal transport geometry, and kernel-based classification.

• Retinal Vessel Segmentation using Mathematical Morphology
  Interpretable multi-scale vessel segmentation pipeline developed from first principles for medical image analysis.

• Graph Neural Networks for Metabolic Pathway Analysis
  From-scratch Graph Convolutional Network for enzyme classification on KEGG biological networks.

Computer Vision & Foundation Models

• GeoAI Crop Mapping with Satellite Foundation Models
  Collaborative project focused on AlphaEarth and Galileo foundation-model embeddings for crop classification and vegetation analysis. I was responsible for working with the foundation-model representations and the downstream classification pipelines.

• Attention U-Net for Sentinel-2 Land-Cover Segmentation
  Custom Attention U-Net for multispectral semantic segmentation (mIoU 0.926).

• Sentinel-2 Unsupervised Land-Cover Labeling
  Explainable unsupervised pseudo-labeling pipeline using Google Earth Engine and ensemble learning.

Explainable AI & Applied Intelligence

• CPFI / CPII — Explainable Decision Intelligence for Construction Project Health (Ongoing)
  Independent research on semantic representation, heterogeneous data alignment, uncertainty-aware reasoning, and interpretable assessment of project health and risk.

• Interpretable TSK Fuzzy Model for Asphalt Performance Prediction
  Transparent fuzzy inference system combining clustering, feature selection, and explainable prediction.

• XAI for MRI Brain Tumor Segmentation (Ongoing)
  Investigation of explainable approaches for MRI-based brain tumor segmentation, with focus on interpretability and reliable decision support.

My research practice emphasizes methodological understanding, independent implementation, and reproducibility.

I prioritize:

• Independent reproduction of research methods
• From-scratch implementation when official code is unavailable
• Experimental design driven by clear research questions
• Careful analysis of representations, assumptions, limitations, and failure modes
• Fully reproducible computational pipelines
• Transparent documentation of methods and results

I aim to move rigorously from:

research question → method → implementation → experiment → evidence

AI Engineer — Freelance
Design and development of machine learning systems and AI-oriented applications.

Software Programmer — Freelance
Software development across Python, systems programming, and applied intelligent systems.

Artificial Intelligence Teaching Assistant
Assignment design, instructional material development, and technical mentoring in AI and machine learning.

Operating Systems Laboratory Instructor
Independent teaching of undergraduate laboratory courses and preparation of complete technical materials.

Programming: Python · C/C++ · SQL
Deep Learning: PyTorch · TensorFlow/Keras · Scikit-learn
Computer Vision: OpenCV · Image Segmentation · Image Reconstruction · Super-Resolution · Mathematical Morphology
Research Methods: Representation Learning · Graph Neural Networks · Graph Kernels · Optimal Transport · Explainable AI · Foundation Models
Tools: Git · Jupyter · LaTeX · Linux · Google Earth Engine · QGIS

My research is directed toward advancing Medical AI, with a particular focus on AI for Cancer.

I am especially interested in:

• MRI-based brain tumor segmentation and analysis
• Explainable and trustworthy models for clinical decision support
• Multimodal and vision-language approaches for medical data
• Foundation models for medical and scientific applications
• Reliable AI systems that can support cancer diagnosis, characterization, and treatment-related research

I am motivated to continue this trajectory through advanced research and aim to contribute as a specialist in AI for cancer, with the long-term goal of pursuing postdoctoral research in this domain.

• GitHub: hannah-fathi
• LinkedIn: hannah-fathi
• Email: hannahfathi99@gmail.com