MacVincent Agha-Oko

MSCS at Stanford

Education

Stanford University

Master of Science, Computer Science

September 2022 - June 2024

Relevant Coursework: Distributed Computing, Machine Learning, Parallel Computing, Deep Learning for Computer Vision, Principles of Robotics.

Artificial Intelligence Specialization

Claflin University

Bachelor of Science, Computer Engineering

August 2018 - May 2022

Relevant Coursework: Algorithms and Data Structures, Operating Systems, Embedded Systems, Linear Algebra

Presidential Scholar | Valedictorian | Summa Cum Laude

Experience

Google

Software Engineering Intern - Assistant Applied Perception

June 2023 - September 2023

  • Deployed an on-device Pathways Language Model (PaLM) variant to improve Google Assistant’s personalized query fulfillment.

  • Implemented a Java Android application that utilizes Android’s Accessibility Service and Media Projection library to capture screen content, the Room Persistence Library for on-device storage, and OpenCL model runners for on-device model inference.

Graduate Researcher

Stanford Vision and Learning Lab

September 2022 - March 2023

  • Worked on a research project focused on implicit scene representations for indoor robot navigation tasks.

  • Trained reinforcement learning policies to perform navigation with the Tiago robot on OmniGibson simulator scenes.

  • Generated a 3D reconstruction dataset consisting of RGB-D images and camera poses for indoor OmniGibson scenes.

Google

Software Engineering Intern - Geo Auto-Moderation

June 2022 - September 2022

  • Developed abuse investigation dashboards used by Google Maps to identify large-scale coordinated activities by bad actors.

  • The dashboards efficiently display network graphs depicting connections between users with similar attributes.

  • Generalized dashboard SQL data source to facilitate custom edge table addition.

  • Made other updates which led to projected annual savings of about $180,000.

Zoom

Software Engineering Intern

January 2022 - April 2022

  • Developed a proof-of-concept real-time collaboration web application using AWS AppSync and Amplify.

  • Maintained internal web applications for the Zoom Apps platform using React and Node.js libraries.

  • Assisted in onboarding and early product research for a new Zoom Chat project.

Google

Software Engineering Intern - Display Ads

May 2021 - August 2021

  • Reduced the deployment time for advertisement filters by decoupling them from serving binaries.

  • Improved metrics monitoring with a regex classifier for OS and browser attributes in ad server requests.

Lyft

Software Engineering Intern - Deploys Team

June 2020 - July 2020

  • Enhanced Kubernetes cluster targeting for Lyft microservice deployments.

  • Improved webpage used to monitor commits to a runtime configuration repository.

  • Automated post-incident batch deployment of backlogged commits to production servers.

Projects

Novel View Synthesis

March 2023

  • Given an input RGB image, we simulate a 3D interactive experience by generating novel viewpoints to account for disocclusion.

  • Builds upon Single Image 3D Photography with Soft Layering and Depth-aware Inpainting (SLIDE) by Jamapani et al., 2021.

  • Unlike SLIDE, we outpaint the input image with a denoising probabilistic diffusion model, use a matting model to separate the background from the foreground image, and construct our meshes with the Open3D library in place of Tensorflow 3D.

Multi-Object Tracker

November 2021

  • Given an input RGB image, we simulate a 3D interactive experience by generating novel viewpoints to account for disocclusion.

  • Builds upon Single Image 3D Photography with Soft Layering and Depth-aware Inpainting (SLIDE) by Jamapani et al., 2021.

  • Unlike SLIDE, we outpaint the input image with a denoising probabilistic diffusion model, use a matting model to separate the background from the foreground image, and construct our meshes with the Open3D library in place of Tensorflow 3D.