vantage/

Shalom Bebebaraseigha

Shalom Bebebaraseigha

●Available for Full-Stack & AI Engineering roles

Shalom Bebebaraseigha

Engineering explainable AI models and the full-stack infrastructure that powers them.

I build systems from the ground up, moving seamlessly from physical network infrastructure to predictive machine learning. Whether I am training classifiers, explaining what they learned through visual interfaces, or shipping local-first and cloud-ready intelligence behind secure APIs, I own the complete technical lifecycle.

Both projects below are built from real data and run locally. The numbers on their pages come straight from the repos, including the unflattering ones.

Class-weighted XGBoost confusion matrix for the oil spill severity classifier, with class order 0 Major, 1 Medium, 2 Minor on both axes. Actual Major cases: 2 predicted Major, 4 Medium, 4 Minor; actual Medium: 2 Major, 38 Medium, 55 Minor; actual Minor: 13 Major, 276 Medium, 2,247 Minor.
Oil spill severity classifierClass-weighted XGBoost

Core competencies

Infrastructure
Network configuration, system setup, hardware support (IT&I placement, NDDC)
Machine learning
XGBoost, scikit-learn, class imbalance, SHAP explainability
Backend & NLP
FastAPI, local LLM pipelines with Ollama, embedding search, Docker
Cloud
Static-first deploys on Cloudflare Pages; containerised services
Motion
After Effects keyframing, typography pacing, audio-synced short-form reels

Featured work

all projects →