Matheus Araujo Email & Phone Number

Montreal, QC, CA

Florianópolis, Santa Catarina, Brazil

Florianópolis, Santa Catarina, Brazil

São Carlos, São Paulo, Brazil

• Research and implementation of novel algorithms for collision detection and response of subdivision surfaces on GPU. More specifically, the main objective is to prevent self-collisions on Physically based Cloth.
• Implemented the system integration between physics engines and geometric modelling libraries, using Object Oriented Programming (OOP), Design Patterns, multithreading and task parallelism.

São Carlos, São Paulo, Brazil

• Teaching assistant of Computer Graphics undergraduate course (SCC0250 and SCC0650).
• Assisting professor for four consecutive semesters (total number of students: 270).
• Helped professor to prepare lecture notes, problem sets and coding assignments.
• Graded problem sets and held office hours, identifying the most common difficulties of the students and providing feedback to the professor.
• Topics tutored include modern graphics pipeline, OpenGL 3/4, GLSL, linear algebra and transformations, rasterization, computational geometry, texture mapping, ray tracing, among others.

Full-time paid research program on computer graphics and geometry processing.

Florianópolis, Santa Catarina, Brazil

• Research and development of CAD and digital sculpting desktop software for the fashion industry, focusing on computer graphics, computer animation, and parallel programming.
• Main languages and technologies: C++17, WebGPU, GPU Compute, CUDA, OpenGL, GLSL, Eigen, Boost, TBB/oneTBB, OpenVDB/NanoVDB, Qt, VTune, NSight.

São Carlos, São Paulo, Brazil

• Research project focusing on geometric deep learning and computer vision applied to point clouds, geometric graphs and triangle meshes.
• Developed a geometry processing pipeline to clean, extract features and apply data augmentation to 3D data scanned from the real-world using Open3D, CGAL and PyMeshLab.
• Implemented point cloud, graph and mesh convolutional networks using Python and PyTorch for tasks such as 3D face recognition, 3D shape classification and retrieval, and sketch-based shape retrieval.
• Implemented siamese and triplet neural networks to combine cross-domain geometric data.
• Achieved 98 % accuracy on 3D face recognition datasets and 80 % mean-average precision on sketch-based shape retrieval datasets.

São Carlos, São Paulo, Brazil

• Participated in weekly meetings with other members to discuss and solve mathematical modelling and scientific computing problems.
• Contributed to the study group of numerical solutions for differential equations and SIR models for propagation of COVID-19.