Computer Vision Proofs of Concept
Rapid prototypes for detection, segmentation, and model-assisted workflows that show whether an AI idea can work in the real world.
AI & Computer Vision Engineer
I turn early computer-vision and AI ideas into testable prototypes and practical workflows. I also build the focused websites that make technical work understandable.

Services
Rapid prototypes for detection, segmentation, and model-assisted workflows that show whether an AI idea can work in the real world.
Practical AI features, evaluation loops, and automation flows for teams that need useful output instead of demos that only look impressive.
Focused pages for founders, creators, and small teams that need a clear offer, a sharp visual presentation, and a path to contact.
About
I study at Howest University of Applied Sciences and build projects across AI, computer vision, and web development. My strongest work is turning early ideas into concrete prototypes that can be shown, tested, and improved.
If you need a landing page, a working proof of concept for a computer-vision project, or a clearer website for your business, I can help you move from a loose idea to something people can understand and contact you about.
Experience
My work sits between experimentation and delivery: I like building the first useful version, testing what holds up, and explaining it clearly to the people who need to use it.
Nov 2025 – Mar 2026
Sweetspot Experts BV
Built AI solutions and internal tools while experimenting with ontologies, semantic chunking, and rapid prototyping.
Jun 2025
EasyPost Belgium
Developed a computer-vision solution for anomaly detection and contributed to a workflow that continued to be reused afterwards.
Jun 2026
Off the Radar
Supported end-to-end event planning, logistics, attendee coordination, and on-site execution for an invitation-only AI gathering in Brussels.
2021 – 2024
Lovia

Founded and bootstrapped Lovia, an app for meeting students, covering UX, beta testing, GDPR, product updates, and MVP validation.
View Lovia on APKPure →Computer vision
Filmed an indoor-soccer match, labeled a custom Roboflow dataset, trained YOLOv8, and ran detection on frames and full video.
Take-home proof of concept · Nimble Studio
Built and deployed a multi-agent energy assistant with structured reasoning workflows. The assistant helps users understand energy billing, book service appointments, and receive personalized energy-saving recommendations using Eve, OpenAI, TypeScript, and custom tool functions.
Portfolio
A few public projects and writeups that show the kind of prototypes and technical experiments I build.

Computer-vision experiment showcased through a LinkedIn project post.

Segmentation workflow shared as a Meta and computer-vision learning project.
Filmed an FC Berdi indoor soccer match and trained YOLOv8 on the footage to detect players and the ball — custom dataset labeled in Roboflow, training tracked with ClearML, detection on single frames and full video.

YOLOv8 medical-imaging proof of concept comparing object detection and segmentation.

Segmentation comparison work connected to the brain-tumor detection article.
Repository Graphs
This graph turns three selected repositories into a browsable map of modules, functions, classes, and relationships. Search a symbol, inspect its source location, and follow connected code instead of opening each repository from scratch.
An interactive map across three selected GitHub repositories, connecting this portfolio site, discovery-report agents, and data visualization work.
Included repos
Contact
Send a short message with the goal, the current state, and what you want to launch or validate. I'll reply by email.
Email: pascal-musa@hotmail.com
LinkedIn: Pascal Musabyimana