I'm a machine learning practitioner interested in explainable AI, small data challenges, and healthcare applications.
My background is in linguistics, and I approach AI through a language and cognition lens, utilizing context through knowledge graphs, ontologies, and semantic web technologies.
Currently based in Ghana, Iβve worked on projects ranging from clinical text classification to building semi-autonomous agents for legal reasoning.
I care about making AI understandable and trustworthy for real-world users β especially in high-stakes fields like medicine.
- Explainable AI for healthcare
- Knowledge-augmented machine learning
- Human-in-the-loop systems
- NLP in clinical and legal domains
-
Medical Screening Pipeline
Used ontology-enhanced ML to reduce manual review time from 6 months to 1 week -
Agentic Legal AI
Developed LLM-integrated agents with tool-use workflows and validation mechanisms -
Churn & Risk Models
Built interpretable models with uncertainty estimates for business and patient decisions
Python, PyTorch, TensorFlow, SHAP, Hugging Face, Azure, REST APIs, SNOMED-CT (medical ontologies), ML Flow, SQL
π Currently Building
GDPR AI Compliance Scorer - Automated regulatory checking for healthcare AI systems Agentic Medical Assistant - Conversational AI that actually understands clinical context Production ML Templates - Because good AI needs good engineering
π± Learning & Exploring
- Advanced RAG architectures & vector databases
- MLOps best practices (currently pursuing certification)
- Swedish SE β preparing for Nordic opportunities
- Explainable AI for high-stakes decisions
- Healthcare AI with limited training data
- Knowledge graph integration with LLMs
- Open-source tools for medical NLP
- M.A. Linguistics (Computational focus), University of Bergen
- Undergraduate in Language & Philosophy
- Certifications: Deep Learning Specialization (in progress), Azure Fundamentals, ML Engineer(Datacamp)
Feel free to reach out β whether you're working on similar problems or just curious.
π§ sammy.okmens@gmail.com
π LinkedIn
-The best AI isn't the most complex β it's the one that solves real problems with the data you actually have.