Alejandro Amat Payá

Research Associate at Carnegie Mellon University

Hi! I'm a Research Associate in the Human Sensing Lab at Carnegie Mellon University's Robotics Institute, working under Prof. Fernando De la Torre. I'm also an incoming MSCS student (Fall 2026), currently taking graduate courses as research staff. I received my dual Bachelor's degree in Electrical Engineering and Computer Science from CFIS-UPC Barcelona.

My work focuses on real-time differentiable rendering, GPU-accelerated 3D vision, human modeling, and neural scene reconstruction. I combine research in graphics and machine learning with high-performance engineering.

Publications

Echoes of the Coliseum: Towards 3D Live Streaming of Sports Events

Junkai Huang*, Saswat Subhajyoti Mallick*, Alejandro Amat, Marc Ruiz Olle, Albert Mosella-Montoro, Bernhard Kerbl, Francisco Vicente Carrasco, Fernando de la Torre

ACM Transactions on Graphics (SIGGRAPH 2025)

paper website video data

Location-based real-time utilization of reconfigurable intelligent surfaces for mmWave integrated communication and sensing

Filip Lemic, Jalal Jalali, Gerard Calvo Bartra, Alejandro Amat, Jakob Struye, Jeroen Famaey, Xavier Costa Perez

IET Advanced Metaverse Wireless Communication Systems, 2025

Projects

Vulkan 3D Gaussian Splatting Renderer

Cross-platform neural rendering engine breaking CUDA vendor lock-in with Vulkan and GLSL compute shaders. Built complete tile-based rasterization, multi-stage radix sort, and ImGui keyframe animation system achieving stable 60 fps on Nvidia 3060 Ti and 24-45 fps on Apple Silicon. It also has a Python library through PyBind bindings.

code

Real-Time 3D Human Avatar System

Leading development of real-time pipelines for capturing and animating digital avatars from monocular video at CMU's Human Sensing Lab. Integrated 3D Gaussian Splatting, SMPL/FLAME body models, pose estimation, segmentation, and canonical optimization with differentiable rendering achieving sub-20-minute training and 15 FPS live streaming for re-animatable avatars.

Real-time Differentiable Rendering with Slang

Built a full C++ rendering system and shader pipeline leveraging Slang’s automatic differentiation. Designed custom optimizers and real-time GPU kernels, achieving 116× speedup over NVIDIA Falcor for inverse texturing and 4× faster performance in 2D Gaussian Splatting, enabling practical real-time inverse rendering.

code

Scotty3D Graphics Pipeline

Implemented core rendering, modeling, and simulation features in C++ for CMU's 15-662 graphics framework. Developed a software rasterizer with perspective-correct interpolation, mipmapping, and supersampling; a mesh editor featuring Loop subdivision, Catmull-Clark surfaces, and isotropic remeshing; and a BVH-accelerated Monte Carlo path tracer with importance sampling, global illumination, physically-based materials (Lambertian, mirror, glass with Fresnel), and HDR environment lighting for photorealistic rendering.

Museum VR Simulation (Unity)

Developed a cross-platform Unity VR simulation for a museum experience in collaboration with researchers from IESE, MIT, and LBS. Implemented gizmo-style controls, gyroscope/accelerometer sensor fusion, crowd simulation, and immersive spatial audio, with Firebase for real-time synchronization and persistent user action storage. Deployed with Google Cardboard for accessible mobile VR.

Web/PC version

High-Performance CUDA Optimization Suite

Collection of GPU-accelerated algorithms with massive performance improvements. Sobel Edge Detection achieving 224733× speedup over NumPy, CUDA-JPEG with 120× speedup vs CPU, and optimized Multi-Scale SSIM for real-time image quality assessment.

code

ZeOS: x86 Operating System Kernel

Extended a didactic OS kernel in C and x86 assembly for UPC's OS course. Implemented system call interface (write, gettime), hardware interrupt handlers (keyboard, clock), and process management (creation, scheduling, context switching, process info tracking) for a multiprocess environment.

code

mmWave Radar & IRS Simulator

Developed a MATLAB simulator for mmWave radar heatmap generation and Intelligent Reflecting Surface (IRS) optimization, enabling multi-user tracking and signal homogeneity analysis for immersive VR environments.

code

HP Construction Verification Library (HP collaboration)

Led development of a Python library supporting the HP SitePrint robot, integrating point cloud segmentation, DBSCAN clustering, and distance verification with Intel RealSense D435i. Achieved <3% error and published on PyPI.

PyPI code demo

Experience

Research Associate

Carnegie Mellon University

Feb 2024 – Present

Conducting research on real-time neural rendering and 3D computer vision at the Human Sensing Lab, bridging graphics, machine learning, and high-performance computing.

AI Researcher

i2CAT Foundation

May 2023 – Dec 2023

Developed real-time 3D imaging algorithms with MIMO FMCW mmWave radar, published 2 papers on intelligent surface optimization.

Unity VR Developer

IESE Business School

Oct 2023 – Apr 2024

Built cross-platform VR application for IESE/MIT/LBS collaboration with Firebase integration and multi-platform deployment.

Big Data Engineer Intern

Mango

May 2022 – Sep 2022

Developed automated data pipelines using Jenkins CI/CD, Airflow, and AWS for large-scale retail analytics.

Education

Visiting Research Student - Bachelor's Thesis

Carnegie Mellon University, Robotics Institute

Feb 2024 – July 2024

Thesis: "Real-Time Differentiable Rendering with Slang" - Developed high-performance differentiable rendering pipelines under supervision of Dr. Fernando de la Torre. Grade: 10/10

Dual Bachelor of Engineering

CFIS-UPC Barcelona

2019 – 2024

Electrical Engineering & Computer Science - Graduated with honors, recognized as 6th best student of 2024 graduating class.