Syed Muhammad A.

I build 3D vision systems: Gaussian splatting, multi-view reconstruction, parametric body models, and the Python/CUDA pipelines that run them in production. Recent work: • 3D Gaussian Splatting for light-sheet microscopy volumes (TU Dresden research project). I built the full training and rendering codebase and deployed it on cloud GPUs. • SMPL-X body-fitting pipeline for an iOS avatar product, with orientation search, mesh validation, and a measurement engine that derives ~50 body measurements from fitted meshes, validated against tape-measure ground truth. • 360° panoramas from sparse, low-overlap images (SuperPoint/SIFT, custom GPU stitching, inpainting) for a virtual-tour product. • Virtual try-on and segmentation pipelines (DeepLabV3+, U2-Net, StyleGAN) for e-commerce clients. • Multi-tenant AI agent platform (forward-deployed engineer): agent workflows, tool integrations, deployment. Stack: Python, PyTorch, NumPy, OpenCV, gsplat, CUDA, camera models and projection math, FastAPI, Docker, GCP/Azure. I'm an MSc student in Computational Modeling & Simulation (Visual Computing) at TU Dresden, based in Germany (CET). Good fit if you need: photo/video to 3D, splat or mesh pipelines, body fitting and measurement, depth/flow/geometry tooling, or synthetic-data processing for ML.