Otavio Rosa

MS CS · USF  ·  Computer Vision & Generative AI

I work on computer vision, generative AI, and synthetic media detection.

My work has included AI-generated image detection, robotic vision systems, and shipping full-stack products.

I'm looking for AI/ML Engineering roles where research depth meets production systems.

Founder of Please Don't Scroll, previously at ClearSet.AI.

01Experience

Founder

Please Don't Scroll

Jun 2026 – Present
  • Designed, built and shipped a Manifest V3 Chrome extension published to the Chrome Web Store, a web dashboard and a Postgres backend, sharing one dependency-free JavaScript core so auth, sync and URL-matching logic live in a single place.
  • Wrote 138 zero-dependency unit tests on Node's built-in runner, covering URL-matching rules and the exact shape of every REST call via a mocked fetch.
  • Automated extension releases with a packaging script that strips dev-only origins, prunes unreferenced files and fails the build on any broken reference, cutting the shipped bundle from 2.45 MB to 83 KB.

Software Engineering Intern

ClearSet.AI

May 2025 – Dec 2025
  • Built 30+ reusable Angular/TypeScript UI components (e.g., data tables, form builders, dashboards) that became the shared base for 6+ product features.
  • Fixed 50+ bugs across the full application stack, from the C# .NET backend to the Angular frontend, implementing unit testing to improve system reliability and contributing to a 15% decrease in critical support tickets.
  • Wrote the SQL migrations for 4 feature rollouts and 10+ production data fixes, all deployed with zero downtime across 20,000+ user records.
02Research

Graduate AI & CV Researcher

Dr. Utkarsh Ojha · University of South Florida

Jan 2026 – Present
  • Co-first author, ICLR 2027 submission: showed real and AI-generated images remain separable under extreme abstraction (edge maps, heavy blur/downsampling) across 10 datasets, suggesting detectors rely on global composition rather than local artifacts.
  • Trained 400+ binary classifiers across 9 architectures (ResNet, DenseNet, SRNet, ViT, and 5 CNN variants) on CelebA, FFHQ, ImageNet, and generated datasets, running 1,000+ GPU-hours of SLURM jobs on USF's HPC cluster.

Undergraduate CV & Robotics Researcher

RARE Lab · University of South Florida

Nov 2025 – Apr 2026
  • Built real-time projector-camera compensation for the Unitree Go2 quadruped (OpenCV, PyTorch, ROS2), reducing projection distortion on irregular surfaces by ~70% at 24 FPS for navigation in simulated disaster environments.
  • Set up an Ubuntu 20.04 VM on Apple Silicon that resolved ROS2 networking issues with the robot's peripherals.
03Skills & Stack

Languages

PythonJavaScriptTypeScriptHTML/CSSC/C++C#SQL

Web

AngularReactNode.js.NETREST APIs

AI / ML

PyTorchOpenCVNumPyscikit-learnMatplotlib

Infra & Tools

GitLinuxPostgreSQLSLURM/HPCROS2Chrome Extensions (MV3)Docker

Spoken Languages

English (Fluent)Portuguese (Native)Spanish (Advanced)Japanese (Lower-intermediate)
04Education

University of South Florida

Current

M.S., Computer Science

GPA: 4.0/4.0

May 2027

University of South Florida

B.S., Computer Science

GPA: 3.9/4.0

Presidential Honors Scholarship, Honors Medallion, 8x Dean's List

May 2026
05Organizations

Community Chair

Google Developer Group

Jan 2025 – May 2025
  • Led planning for HackUSF 2025 (300+ participants), managing 10+ organizers, timeline and logistics, and budget of $5,000.