cv
Basics
Name | Nemanja Bartolovic |
Label | Software Engineer |
nemanjabartolovic@gmail.com | |
Url | https://fallenshard.github.io |
Work
-
2021.08 - Present Staff Software Engineer
Meta
real-time rendering, neural rendering, performance optimizations, cloth simulation
- Spearheading the initiative and development for a unified photorealistic real-time renderer for environments and digital avatars in VR via modern neural rendering approaches, like Gaussian splatting. Shipped Meta Horizon Hyperscape, a demo app that showcases photorealistic environments on Quest devices.
- Reduced training costs in multiple millions of dollars per year by optimizing CUDA kernels and training pipelines for large-scale ML efforts in 3D reconstruction of digital avatars.
- Scaled high-fidelity cloth simulations and rendering for digital humans. Led development of various cloth simulation components, both on CPU (SIMD) and GPU (CUDA), as well as real-time neural ML deformers (Vulkan, Torch) that run on VR devices.
- Created several cross-platform libraries related to computer graphics used by dozens of internal teams in diverse projects (acceleration structures, Vulkan abstraction layer, GPU tracing, Gaussian splatting runtime).
-
2020.09 - 2021.08 Software Engineer
Daedalean AG
real-time rendering, machine learning, flight simulations
- Worked on reliable autonomous piloting systems for aircraft through vision-based methods.
- Developed a near real-time cloud-based rendering and simulation service to generate large-scale multi-view aerial synthetic data for multiple internal teams.
- Improved the detect-and-avoid system by scaling its infrastructure for training models, sophisticated data augmentation (TensorFlow) and safety-critical testing.
-
2019.05 - 2020.08 Research Assistant
ETH Zürich
scientific visualization, flow analysis, dimensionality reduction, machine learning
- Developed novel techniques for dimensionality reduction, topology, feature extraction and visualization of high-dimensional flows, gravitational and general dynamical systems. Primary methods of approach included nonlinear optimization, machine learning and variational calculus.
-
2017.06 - 2019.05 Rendering Engineer
Kapanu AG
real-time rendering, software architecture, geometry processing, inverse rendering
- Shipped IvoSmile, a mobile AR try-on tool for dentists that aids in communicating dental prosthetic treatments.
- Developed an efficient mobile-first software architecture (iOS, Android), a cross-platform real-time rendering system (OpenGL), optimized performance and developed several geometry processing algorithms for dentures. Conducted R\&D on rendering and machine learning as well as physical lab design for capturing dental materials
Education
-
2015.09 - 2017.04 Zurich, Switzerland
-
2011.10 - 2015.07 Nis, Serbia
Publications
-
2025 SqueezeMe: Mobile-Ready Distillation of Gaussian Full-Body Avatars
SIGGRAPH 2025
Co-developed a method to animate and render multiple digital human avatars in VR.
-
2025 Hardware-Rasterized Ray-Based Gaussian Splatting
CVPR 2025
Co-developed a novel rasterization-based Gaussian splatting algorithm that minimizes projection errors in high field-of-view settings like VR rendering.
-
2020 Phase Space Projection of Dynamical Systems
EuroVis 2020
Developed a flow and curvature-preserving dimensionality reduction technique for trajectories in high-dimensional dynamical system phase spaces. One of the best papers with Honorable Mention Award.
-
2017 PhysicsForests: Real-Time Fluid Simulation Using Machine Learning
SIGGRAPH 2017
Implemented rendering techniques used to visualize fluid simulations which were initially developed during my Master's thesis, as well as improvements to the core simulation framework.
Languages
English | |
Fluent |
Serbian | |
Native Speaker |
German | |
Basic |