Researcher at CUET
I am a self-driven researcher transitioning from a mechanical engineering background into advanced computer science, with a strong focus on data science, computer vision, and physics-informed machine learning. I aim to build mathematically grounded models rather than relying purely on empirical approaches. My primary research interests lie in combining deep learning with physical principles—especially through Physics-Informed Neural Networks (PINNs). I am actively exploring how to integrate physics constraints into domains like 2D/3D human pose estimation, airfoil flow prediction, and broader scientific modeling. I am particularly interested in developing architectures beyond standard pipelines, such as hybrid CNN–PINN systems, regime-aware models, and latent-space optimization techniques. I am strongly motivated toward high-impact research and aim to publish in top-tier (Q1) journals. My focus is on mathematical correctness, interpretability, and novel contributions, including areas like superposition-based latent encoding, uncertainty-aware Bayesian methods, and geometry-to-physics prediction pipelines. I am not just applying existing methods—I am working toward developing new frameworks. Technically, I am building skills in: PyTorch-based deep learning systems Scientific machine learning and PDE-constrained modeling Computer vision (pose estimation, feature extraction) Working with real-world datasets such as Human3.6M, MPII, and CFD/airfoil data Research-level experimentation and model evaluation At the same time, I am navigating real-world challenges, including financial pressure, career uncertainty, and transitioning across disciplines. My immediate goal is to secure a remote opportunity while steadily building toward a long-term research career.