Available for opportunities
Full-Stack Developer & AI/ML Engineer
I design and build whatever I can imagine. Building robust digital products and intelligent systems from Addis Ababa.
currently_building: intelligent_systems
Currently
AI/ML Intern
iCog Labs
Addis Ababa, Ethiopia
Web systems that work.
AI systems that understand.
Built from first principles.
I'm a full-stack developer and AI/ML engineer based in Addis Ababa, Ethiopia. My core stack is React, Laravel, Tailwind, and Express, with a strong focus on backend systems and data-driven applications.
Alongside web development, I work on applied AI/ML — building retrieval-augmented generation (RAG) pipelines, computer vision models, and neural networks implemented from first principles rather than just calling high-level APIs.
Currently an AI/ML intern at iCog Labs, working on applied research spanning RAG systems, predictive coding networks, object detection, category theory, and probabilistic logic networks (PLN).
RAG evaluation · Predictive coding · Computer vision · Production web apps
Current
iCog Labs
Applied AI/ML research and implementation, including retrieval-augmented generation (RAG) systems built with LangChain and evaluated with Ragas, predictive coding network architectures trained without standard backpropagation, YOLOv8-based object detection, and theoretical work in category theory and probabilistic logic networks (PLN).
HiLCoE School of Computer Science & Technology
Scientific Computing with Python
freeCodeCamp
Machine Learning Specialization
Coursera (DeepLearning.AI / Stanford)
Deep Learning Specialization
Coursera (DeepLearning.AI)
Featured project
A restaurant web application designed to make the dining experience more accessible through a focused, production-ready interface.
Case study: Explore the implementation, tradeoffs, and lessons learned in the repository.
Open projectFeatured project
A personal protective equipment (PPE) detection pipeline using YOLOv8, trained on a Roboflow dataset to detect workers and verify required safety gear compliance on-site.
Case study: Explore the implementation, tradeoffs, and lessons learned in the repository.
Open projectFeatured project
A neural network model classifying traffic accident types using cleaned and engineered features from a large real-world dataset, with careful preprocessing and validation to prevent overfitting.
Case study: Explore the implementation, tradeoffs, and lessons learned in the repository.
Open projectLive from GitHub
Latest repositories load from GitHub when available.
RAG / EVALUATION
What I learned measuring retrieval quality, faithfulness, and answer relevance with Ragas.
Draft exampleDEEP LEARNING
Implementing cross-entropy and gradient descent to understand what high-level frameworks abstract away.
Draft exampleAI RESEARCH
Exploring learning rules that minimize prediction error without relying on standard backpropagation.
Draft exampleCOMPUTER VISION
Dataset choices, augmentation, and evaluation decisions behind a YOLOv8 safety-gear detector.
Draft example06 — Start a conversation
Available for full-time roles, freelance projects, and thoughtful collaborations.
Based in Addis Ababa, Ethiopia. Open to remote work. I typically respond within 1–2 business days.
yaredmengiste01@gmail.com