Gustavo Silvera

gustavo [at] silvera [dot] cloud


Hello there   Welcome! I enjoy working in the intersection between computer systems, graphics, and AI, especially in the field of autonomous driving and simulation.


I'm a graduate from the CMU School of Computer Science with a major in artificial intelligence (BSAI) and concentration (minor) in computer graphics. I graduated in May 2023.


GitHub  /  G Scholar  /  Resume  /  Linkedin  /  Email


TL;DR:
- I like computer systems, graphics, games, and AI.
- I am interested in autonomous and assistive driving technology.
- I am also interested in photorealistic rendering and realistic simulation.
- I have interned at (formerly) Facebook in infra and Meta in XR holograms.
- Graduated from CMU SCS in May 2023.
- Reach me via email: gustavo@silvera.cloud



Academic Experience (Research)

Starting in 2020 I joined the CMU Robotics Institute's (RI) HARP and TBD labs, led by Dr. Henny Admoni and Dr. Aaron Steinfeld respectively. I worked closely with PhD student Abhijat Biswas on designing and building the backend core simulator functionality for a new research project: SocNavBench [1]. This project introducd me to academic writing (published at the ACM THRI Journal 2022) and research engineering.

I then continued with HARP lab to build a eye-tracking VR driving simulator for Human Robot Interactions (HRI) research in the form of DReyeVR [2] (read "driver"). This project extends the popular open source CARLA simulator for autonomous vehicles, but focuses on the "human in the loop" aspect of driving research. In addition to implementing the core human-immersion enhancements (VR, audio, haptics, physical driving hardware, etc.) and extending CARLA's research features (serializer, recorder/replayer, PythonAPI, etc.), I am first author on a workshop paper submitted to ACM/IEEE HRI '22 and I'm also actively maintaining the released project on GitHub. Happy to take any questions!





Publications:

  1. A. Biswas, A. Wang, G. Silvera, A. Steinfeld, and H. Admoni. 2022. SocNavBench: A Simulation Testing Framework for Evaluating Social Navigation.

  2. G. Silvera, A. Biswas, and H. Admoni. 2022. DReyeVR: Democratizing Virtual Reality Driving Simulation for Behavioural & Interaction Research.



Academic Experience (Coursework)

I have graduated from CMU SCS with a BSAI (Bachelors in AI) and concentration (minor) in computer graphics in May 2023. Throughout my undergraduate I've taken several courses related to CS, graphics, systems, and AI alongside the standard 2019 curriculum for SCS majors. Here are some of my favourites:


Graphics Courses:

  • 15-462: Computer Graphics

  • 15-463: Computational Photography*1

  • 15-466: Computer Game Programming

  • 15-468: Physics Based Rendering 2

  • 15-469: Visual Computing Systems

AI Courses:

  • 10-315: Machine Learning

  • 11-485: Deep Learning

  • 15-281: AI Representation

  • 15-482: Autonomous Agents

  • 16-385: Computer Vision

Systems Courses:

  • 15-213: Intro to Computer Systems

  • 15-418: Parallel Computer Architecture

  • 15-740: (Grad) Computer Architecture

(*) = Was a teaching assistant for Dr. Ioannis Gkioulekas. Hosted office hours & graded assignments.

(1) = Won the final project competition with minimal-timelens, got to keep the course Camera: Nikon D3500

(2) = Won the final project competition with Practical-PathGuiding, got a fully funded trip to SIGGRAPH 2023


Work Experience (Internships)

I've previously worked at Facebook/Meta during Summer 2021 & 2022. My first internship (2021) was on the Anomaly Detection Infrastructure team and my second (2022) was on XR Holograms Creation.

In my infra internship I build monitoring scripts in python to detect, diagnose, and remediate detection models according to various custom SLIs (Service Level Indicators) ranging from staleness to quality performance. These efforts spanned over 200,000 active production models and alleviated tedious labor from the On-Call engineers who would otherwise need to manually verify and prune or update.

In my XR Holograms internship I built internal tooling to better visualize 3D-volumetric-temporal data (think "3D video call"). This enabled engineers unfamiliar with game engines such as Unity to visualize and debug their data in a lightweight and simple manner. I also explored and presented potential avenues for realistic representation using photorealistic "Codec" avatars attached to stylized horizon-style avatars. I also contributed [REDACTED] to a MZ (yes, him) demo showcasing future possibilities for realistic representations in the Metaverse.


Cool Projects Gallery

Click on any to go to their Github repos


(Most of the thumbnails play a short video when hovered)


DReyeVR

An immersive VR driving simulator built on CARLA for HRI research with eye tracking.

SocNavBench

A benchmarking framework for social navigation algorithms in crowded environments

Practical Pathguiding

Implementation of Practical Path Guiding in the forward path tracer for Mitsuba3

Minimal Timelens

An event-camera based video interpolation project based on TimeLens using pytorch.

GL FovRender

OpenGL project demonstrating fixed foveated rendering on fragment shaders.

Project Athena

A 3D OpenGL space game with realistic orbital mechanics to redirect a rogue asteroid.

Parallel Boids

Boid simulation and parallel performance experiments and analysis with OpenMP/Cuda.

Sporshmallow

A 2D fighter platform arcade game from CMU's Game Creation Society in Unity

Graphical CNN

A visualization of a Convolutional Neural Network using C++ and OpenFrameworks2D

VEX Robotics Simulation

Top-down graphical simulation for the VEX Robotics 2017-18 competition “In the Zone”.

Tennis Simulation

Virtualized computer simulation of a tennis match with adaptive computer player.

FourierDraw

Drawing arbitrary functions (image outlines) with discrete Fourier series approximations

Pi Collider

Graphical visualization of idealized block collisions to compute Pi with physics.