|
omar@workstation:~ (zsh)
arch: x86_64 | session: active
omar@workstation:~ (zsh)
$ neofetch --profile omar-rehan
System Summary
--------------------------------------------------------------------------------------------------
Developer : Omar Rehan
Focus : AI Engineering • Computer Vision • Full-Stack Architecture
Location : Dubai, United Arab Emirates (Open to Global Relocation & Remote)
Education : B.Sc. in Computer Science
Website : https://omar-rehan.vercel.app
Flagship : medtrace (Longitudinal 3D Brain-MRI SegResNet & Lesion Matching)
Specialties
--------------------------------------------------------------------------------------------------
- Deep Learning & 3D Medical Vision (PyTorch, TensorFlow, MONAI, Grad-CAM)
- Vision-Language Model Fine-Tuning (Qwen2.5-VL, Llama 3.2 Vision, LoRA, Unsloth)
- Full-Stack Web Development (Next.js, React 19, TypeScript, Tailwind, FastAPI)
- Cloud, Database & Auth (Firebase, Supabase, PostgreSQL, Docker, Vercel)
- Business Intelligence & Analytics (Power BI, DAX, Streamlit, Plotly, Pandas)
Philosophy
--------------------------------------------------------------------------------------------------
"Engineering intelligent, explainable, and scalable systems that solve high-stakes challenges." |
Production systems, multimodal research architectures, and scalable full-stack platforms.
|
Longitudinal brain-MRI analysis system featuring 3D SegResNet segmentation, affine/deformable registration, spatial lesion matching across multi-visit scans, uncertainty quantification, and Grad-CAM explainability.
|
Unified developer operations dashboard synchronizing GitHub, Vercel, and Firebase metrics in real time. Features automated uptime health alerts, audit logging, and an integrated LLM assistant for instant log diagnosis.
|
|
Domain-adapted Vision-Language Model (VLM) fine-tuned with parameter-efficient LoRA on astronomical spectra and deep-sky imaging. Achieves state-of-the-art visual question answering on celestial structures.
|
End-to-end commercial analytics platform integrating a relational SQL warehouse, RFM customer segmentation, churn probability modeling, and predictive revenue forecasting with confidence intervals.
|
Repository index categorized by domain and technical specialization. Click any section to expand.
AI Systems, Multimodal Foundation Models & Medical Imaging
| Repository | Focus & Capabilities | Tech Stack |
|---|---|---|
| medtrace | Longitudinal 3D brain-MRI segmentation, registration, and lesion matching. | PyTorch MONAI FastAPI |
| Fine-tuning Qwen2.5-VL | 4-bit quantized LoRA fine-tuning for domain-adapted astronomical VLM. | HuggingFace Transformers Gradio |
| Llama 3.2 11B Vision | Memory-efficient multimodal instruction tuning with Unsloth on GPUs. | Unsloth PyTorch LoRA |
| YOLOv8-OBB SAR Detection | Radar imagery vehicle detection with oriented bounding boxes & DOTA format. | YOLOv8 OpenCV Gradio |
| Brain Tumor Grad-CAM | Transfer learning for medical classification with visual saliency heatmaps. | TensorFlow Keras Grad-CAM |
| COVID-19 Chest X-Ray | ResNet50V2 deep neural network with robust data augmentation pipelines. | ResNet50V2 Python OpenCV |
Full-Stack Applications & Cloud Platforms
| Repository | Focus & Capabilities | Tech Stack |
|---|---|---|
| RehanPulse | Real-time DevOps monitor with GitHub, Vercel, Firebase & embedded AI copilot. | Next.js TypeScript Firebase |
| My Portfolio | Production portfolio featuring Supabase auth, CMS control panel & dark mode. | Next.js React Tailwind Supabase |
| TerraExplorer | Interactive country exploration engine with geospatial maps & Gemini facts. | React TailwindCSS Gemini API |
| Finance Tracker | Personal expense tracking, budgets, financial targets & real-time sync. | React Firebase TailwindCSS |
| Profile README | Automated GitHub Actions profile with 100% uptime analytics & snake workflow. | GitHub Actions SVG Markdown |
Data Analytics, Business Intelligence & Data Science
| Repository | Focus & Capabilities | Tech Stack |
|---|---|---|
| E-commerce Sales Intelligence | PostgreSQL data warehouse, RFM segmentation, forecasting & Streamlit UI. | SQL Power BI Streamlit Plotly |
| HR Analytics Dashboard | Executive Power BI dashboard with 50 DAX measures & attrition risk scores. | Power BI DAX Data Modeling |
| Data Analysis Portfolio | Comprehensive analytics: Tableau, Power BI, MongoDB, clustering & EDA. | Tableau Power BI MongoDB Python |
Computer Vision & Deep Learning Lab
| Repository | Focus & Capabilities | Tech Stack |
|---|---|---|
| Cats vs Dogs InceptionV3 | High-accuracy transfer learning achieving ~99% accuracy with tf.data. | InceptionV3 TensorFlow Keras |
| Image Classification & QA | Preprocessing pipeline detecting corrupted, duplicate, and bright artifacts. | TensorFlow Image Processing |
| CNN Autoencoder Denoising | U-Net convolutional autoencoder for removing high-frequency noise from images. | PyTorch Autoencoders CNN |
| CNN vs Transformer vs Xception | Empirical benchmark of DeiT-Tiny Vision Transformers vs classic ConvNets. | PyTorch Vision Transformers |
| Sports Ball Classifier | Two-stage training with balanced datasets and FastAPI deployment. | FastAPI Keras InceptionV3 |
| Audio Mel-Spectrogram CNNs | Audio signal transformation to 2D Mel spectrograms for CNN classification. | Librosa PyTorch Audio DSP |
|
|
|
|
|
|
