Data Scientist with a passion for building robust, expressive deep learning models. My research focuses on uncertainty-aware models and their applications across different domains.

🎓 M.Sc. in Electrical Engineering (Shahrood University of Technology)
🔬 Research Interests: Probabilistic Deep Learning • Signal & Image Processing • Computational Neuroscience • Autonomous

• QiVC-Net (2025)
  A quantum-inspired variational convolutional layer that injects structured, geometry-preserving uncertainty via unitary-like rotations, without extra parameters. Validated on biosignals, but framework is domain-agnostic.

• Dynamic Memory Fusion (2024)
  An adaptive multi-loss framework that dynamically reweights losses in real time using memory of past performance. Applied to medical segmentation but generalizable to any multi-objective task.

• Probabilistic Crypto Forecasting (2023)
  Uncertainty-aware price prediction using transfer learning, showcasing DL robustness in non-stationary financial time series.

• DeepFaceAR (2020)
  A real-time deep face recognition system with augmented reality overlay, combining CNN-based identity verification with AR rendering to display personal information. Demonstrates early mastery of embedded vision pipelines and human-centered AI interfaces.

A selection of earlier implementations reflecting versatility across languages, domains, and abstraction levels:

• Deep learning course with TensorFlow and Keras (2018) — Link
• Image panoramic stitching using OpenCV in C++, Python, and MATLAB (2018) — Link
• Android Image Processing App with OpenCV (2019) — Link
• High-Resolution Video Processing Using OpenCL and OpenCV (2019) — Link
• Implementation of KMeans algorithm from scratch in Python (2021) — Link
• Implementation of an autoencoder using the recirculation algorithm from scratch (2022) — Link
• Aparat video viewer bot using Python (2022) — Link

🌍 My philosophy: Build deep learning methods that are mathematically grounded, computationally efficient, and adaptable across domains, not just tuned for one dataset.
📬 Open to collaboration on theory-driven or applied DL projects.
✉️ Email: [email protected] | 🌐 Personal Site