I am a Physicist and Quantitative Researcher specializing in stochastic dynamics, morphological evolution, and scientific machine learning. My work bridges physics-based mathematical modeling and modern data-driven architectures to analyze high-dimensional systems, optimize complex networks, and build scalable modeling tools.

🔬 Core Focus & Expertise:

• Stochastic Modeling & Non-linear Dynamics: Simulating non-equilibrium systems, Langevin dynamics, and high-dimensional time-series data.
• Scientific Machine Learning: Integrating statistical physics principles with deep learning (Graph Neural Networks, Optimization Algorithms) for pattern recognition and signal recovery.
• Computational Engineering & Analytics: Developing automated pipelines, matrix operations, and interactive dashboards for complex data visualization.

📊 Quantitative Application: Passionate about translating advanced physical and mathematical models into real-world applications across Quantitative Risk (VaR/stress testing), Fraud & Anomaly Detection, and FinTech Credit Analytics.

🤝 Open to Collaborations: Quant Research, Financial Engineering, Scientific ML, and Applied Stochastic Modeling.

📫 Contact: hisaylama@gmail.com

Here are some of my notable projects:

1. Credit Risk Analytics & Real-Time Scoring (LendingClub Case Study) - Engineered an end-to-end Machine Learning pipeline for consumer credit default risk assessment. Implemented automated feature engineering, model training, and a low-latency API/web application for live credit scoring. [FinTech / Credit Risk & Machine Learning]

2. Graph-Based Pattern & Anomaly Detection from Images - Mapped complex image structures to spatial graphs and topological networks to detect structural defects and anomalous junctions. Applicable to complex network analytics, fraud detection, and transactional anomaly signals. [Quantitative Image Processing]

3. Interactive Analytics App for High-Dimensional Data (Chemical Fingerprint) - Developed an interactive GUI to filter, slice, and visualize multi-dimensional feature matrices in real-time. Designed to streamline EDA and complex dashboarding for large-scale dataset analysis. [Data Engineering / Interactive Dashboards]

4. Optimization-Driven Signal Reconstruction (Phase retreival algorithm) - Implemented iterative inverse-problem algorithms (Gerchberg–Saxton phase retrieval) to reconstruct noisy or incomplete optical signals. Methodologies directly extend to financial time-series smoothing, missing-data imputation, and matrix completion. [Signal Processing / Inverse Problems]

5. Stochastic Monte Carlo & Langevin Simulation - Architected agent-based stochastic simulation engines using Langevin dynamics to model non-linear particle interactions. Framework translates to simulating extreme market scenarios, Value-at-Risk (VaR) estimation, and liquidity stress testing. [Quantitative Risk/Stochastic Modelling]

6. Dynamic Phase Transitions in Non-Equilibrium Systems (Nonequillibrium Physics) Engineered an automated MATLAB $\rightarrow$ Python computer vision pipeline (denoising, segmentation, feature extraction) to convert raw microscopy video into quantitative dynamic signals; detected jamming transitions in complex soft matter. Published in PNAS Nexus. Code/Data · [Scientific Machine Learning / Computer Vision]
   Media coverage: ScienceDaily · Phys.org · Bioengineer.org · University of Tokyo

• LinkedIn

Thank you for visiting my profile! If you share similar interests or have exciting collaboration opportunities, feel free to get in touch.