M. Miyad Islam Nion

Electrical & Electronic Engineer | Machine Learning Developer Focus Areas: • Graph Neural Networks (GCN, GAT) • Explainable AI (SHAP) & Anomaly Detection • Smart Grid & Cyber-Physical System Security • Time-Series Forecasting Stack: Python, PyTorch, PyG, Scikit-Learn, Django, FastAPI --- ### 🔬 Research & Technical Interests - **Graph Neural Networks:** Node classification, topological analysis, and link prediction with GCNs and GATs. - **Explainable AI (XAI):** Post-hoc interpretability and feature attribution using SHAP and Integrated Gradients. - **Cyber-Physical Security:** Anomaly detection, fault localization, and cyber-attack defense in smart grids. - **Time-Series & Forecasting:** Sequential modeling for predictive infrastructure and telemetry. --- ### 🛠️ ML & Engineering Stack - **Frameworks:** PyTorch, PyTorch Geometric (PyG), Scikit-Learn, Optuna, Hugging Face Hub - **Analysis & Explainability:** SHAP, NetworkX, NumPy, Pandas, Matplotlib / Seaborn - **Backend & Deployment:** Python, Django, FastAPI, Docker, REST APIs - **Hardware Integration:** Embedded microcontrollers, IoT sensor integration, power system simulation --- ### 🚀 What I Share on Hugging Face - **Models:** Specialized graph architectures and lightweight tabular/time-series anomaly detection checkpoints. - **Spaces:** Interactive Gradio demos exploring GNN predictions, SHAP value visualizations, and signal classification. - **Datasets:** Cleaned benchmark telemetry and synthetic intrusion detection datasets.