Hi, I'm Sam
Hi, I’m Sam.I build ML systems for medical and legal applications, focusing on making AI work with small datasets while remaining explainable to domain experts.My approach combines medical knowledge (ontologies, clinical guidelines) with machine learning to address healthcare's data scarcity challenge. Recent work includes systematic review automation that reduced screening time from 6 months to 1 week.
Research Background
MA, Computational Linguistics — University of Bergen (2024)• Reduced systematic-review screening from months to roughly a week through calibrated, reviewer-ready outputs.• Experience teaching and mentoring graduate students and conducting technical training sessions.• Oslo legal-AI startup experience (Innovation Norway support).How I work: ontology-enhanced ML, leakage-safe evaluation, uncertainty & calibration, and iterative human-in-the-loop reviews, that result in reproducible pipelines with versioned data/code.
Projects
Featured Projects1. Medical intervention text triage (systematic reviews)Classifier augmented with medical ontologies (e.g., SNOMED) to pre-screen literature.
• Outcome: screening time cut from ~6 months to ~1 week in pilot settings.• Assurance: held-out and external-cohort checks; reviewer-level agreement; model card and audit log.• Approach: start simple; add complexity only where error analysis shows value.2. Customer analytics with uncertainty (selected non-medical)Churn prediction with explainability and confidence intervals.
• Outcome: interpretable drivers of churn; decisions supported with confidence.• Assurance: leakage checks, stable splits, and calibration.3. Legal AI System (Oslo Startup)Production ML for regulatory text analysis; supported by Innovation Norway and Microsoft.
• Outcome: 25+ documents processed weekly with explainable rationales.• Assurance: traceable decisions, versioned models, and governance hooks.4. Human–AI creative analysis (research) Studied 1,298+ interactions in a multimodal community to understand prompting and model behavior.
• Outcome: insights into prompt–image patterns; methodology shared for reproducibility.
Current Projects
GDPR Healthcare AI Compliance Scorer
An automated, regulator-ready assessment tool for GDPR/EU-AI-Act controls in healthcare.• Designing as modular “packs” (GDPR, AI Act, MDR, Integrity/anti-corruption).• Produces cited findings, evidence snippets, and early-warning checks—before issues become problems.
Beyond the CodeOutside work you’ll find me with jazz, a good kitchen session, or some sci-fi. I’ve lived in Ghana and Norway, with time in China and throughout Europe—a valuable perspective when building inclusive, globally aware systems. I enjoy making complex tech accessible, from teaching Python to medical researchers to helping technicians make sense of diagnostics.
Currently learning Norwegian, plus some German and Japanese.
Contact
Let's Connect📧 [email protected]
💼 linkedin.com/in/sammens
🔗 github.com/SamInMotion
📍 Ghana (Open to Nordic relocation)Interests: production-oriented roles in compliance, healthcare AI, and research on safe AI for critical applications
Technical blog coming soon - insights on explainable AI and healthcare compliance