Profile Picture

Erjon Musa

AI/ML Engineer | Computer Engineering

Computer Engineering graduate from Queen's University with a focus on computer vision, deep learning, and embedded AI. I like building systems that take ML research and make it work in the real world, whether that's running object detection on edge devices or working with vision-language models.

About Me

Hey, I'm Erjon! I'm a Computer Engineering graduate from Queen's University. I love tackling hard technical problems and building things that matter, whether that means working close to the bare metal or building out scalable full-stack applications.

One project I'm particularly proud of is MyEyes, an assistive AI wearable I architected for the visually impaired. Getting a YOLOv8 object detection model to run low-latency, real-time inference on an Android device, paired with an ESP32 microcontroller for video streaming, was a huge challenge. It pushed me to deeply optimize hardware pipelines and multi-threaded processing. Seeing it actually work reliably on the edge was an incredibly rewarding experience that solidified my love for engineering.

What I Am Looking For

I'm looking for new-grad software engineering roles where I can make a real impact. Because my background spans embedded systems, AI/ML, and full-stack development, I thrive in environments where I can work across the stack. Whether it's building intelligent backend systems, shipping user-facing features, or optimizing algorithms, I'm genuinely open to it. I care most about solving meaningful problems alongside a strong team I can learn from.

Personal Interests

When I'm not coding, you'll probably find me out with my camera doing street or landscape photography. I also love cycling, playing basketball, squash, and frisbee. Cooking and music are big parts of my downtime too.

Technical Skills

AI & Machine Learning

PyTorch
PyTorch
TensorFlow
TensorFlow
scikit-learn
scikit-learn
NumPy
NumPy
Pandas
Pandas
Matplotlib
Matplotlib
Jupyter
Jupyter
Google Colab
Google Colab

Programming Languages

Python
Python
C
C
C++
C++
Java
Java
C#
C#
Kotlin
Kotlin
SQL
SQL

Web & Frontend

JavaScript
JavaScript
TypeScript
TypeScript
React
React
Next.js
Next.js
Tailwind CSS
Tailwind CSS
HTML5
HTML5
CSS3
CSS3

Tools & Platforms

Git
Git
Figma
Figma
Azure DevOps
Azure DevOps
Power Platform
Power Platform
Arduino
Arduino
VS Code
VS Code
Claude Code
Claude Code
Google Antigravity
Google Antigravity
Android Studio
Android Studio

Experience

Government of Ontario

UX/UI Design Team Lead & Frontend Developer

Government of Ontario

September 2024 - August 2025

Built a portal on Microsoft Power Platform and Dynamics 365, replacing old PDF workflows with a proper web app. Led a team of 3 co-op students, ran Agile sprints, and made sure everything met Ontario Design Standards for accessibility.

JavaScript iconJavaScriptC# iconC#Figma iconFigmaPower Platform iconPower PlatformAzure DevOps iconAzure DevOps

Projects

MyEyes AI Glasses

MyEyes AI Glasses

2026

Award-winning capstone project. Smart glasses that help visually impaired users navigate safely using real-time object detection with YOLOv8n on an ESP32-S3, plus OCR for reading signs and text out loud. Comes with an Android companion app built in Kotlin.

YOLOv8TensorFlow Lite iconTensorFlow LiteKotlin iconKotlinPython iconPython

CLIP Fine-Tuning on MS COCO

2025

Took OpenAI's CLIP model (ViT-B/32) and fine-tuned it on MS COCO 2014 to get better image-text matching. Built the full contrastive learning pipeline from scratch and tested how well it transfers to new tasks without extra training.

PyTorch iconPyTorchCLIPVision TransformersPython iconPython

Semantic Segmentation with Knowledge Distillation

2024

Used knowledge distillation to train a small, fast MobileNetV3-ASPP model to segment images almost as well as a much larger teacher network. Tested on PASCAL VOC 2012 with solid results despite the huge size reduction.

PyTorch iconPyTorchMobileNetV3Knowledge DistillationPython iconPython
SnoutNet: Pet Nose Localization

SnoutNet: Pet Nose Localization

2024

Trained three different CNN architectures (a custom SnoutNet, AlexNet, and VGG16) to find pet noses in photos. Got the error rate under 2% on the Oxford-IIIT Pet dataset.

PyTorch iconPyTorchCNNsComputer VisionPython iconPython

Education

Queen's University

BASc Computer Engineering · 2021 – 2026

Excellence Entrance Scholarship

Year 5

2025 – 2026

Capstone: MyEyes AI Glasses

Capstone

Embedded AI | YOLOv8 | TensorFlow Lite | ESP32-S3 | 3rd Place, Capstone Competition

Course

Computer Vision & Deep Learning

CNNs | Object Detection | Semantic Segmentation | Knowledge Distillation

Machine Learning & Deep Learning

Course

Neural Networks | Optimization | Generalization | PyTorch

Course

Artificial Intelligence

Search | Planning | Probabilistic Reasoning | Decision Making

Distributed Systems

Course

Distributed Architectures | Consensus | Replication | Fault Tolerance

Club

Queen's Hyper Computing Club

High-Performance Computing | Parallel Systems

Year 4

2024 – 2025
Co-op

Co-op — Government of Ontario

UX/UI Design Lead & Frontend Developer | Power Platform | Dynamics 365

Contact

Feel free to reach out!
I'm always open to connecting.