Emergency Physician → Clinical Systems Engineer

I’m an emergency physician who builds software and automation tools

I fix things. Sometimes with code. Sometimes with adrenaline. I focus on:

• Small, auditable tools (FastAPI, SQLite, scikit-learn) that fit into existing workflows.
• Safety, transparency, and rollback paths—patient care always comes first.
• Local-first design (no cloud dependencies, no unnecessary data sharing).
• Advice-only systems—augmenting clinical work, never replacing judgment.

"I don’t build AI for healthcare. I build tools for healthcare professionals—where it adds real value."

Python • scikit-learn • FastAPI • SQLite • MIMIC-IV

• CatBoost Models: provides real-time, interpretable risk predictions to support clinician decision-making.
• SPC Monitoring: Shewhart/EWMA for real-time performance tracking (review prompts only)
• Decoupled Architecture: Prediction pipeline + separate LLM assistant service
• Local-Only: No cloud, no stored data - in-memory processing with JSONL audit logs

Key Features: ✅ Advice-only outputs (human approval required) ✅ Full audit trails for all system interactions ✅ Configurable probability thresholds ✅ Monthly SPC baseline updates

Status: Core pipeline functional

Python/FastAPI → Windows .exe (PyInstaller)

A local rules app for ICD-10 eligibility checks:

• Runs offline, no data stored, ships with installer scripts.
• Status: Core engine done; pilot testing next.

Hugging Face T5-small • DataCollator • Beam Search

Debugging-driven fine-tuning:

• Solved a "True"-only bug via systematic isolation (task prefixing, pipeline cleaning).
• Lesson: Debugging ML like a physician—rule out causes one by one.
• Notebook: Kaggle

I use the term Clinical Systems Engineer because:

• "Clinical" means starting with real healthcare problems.
• "Systems" means focusing on entire workflows, not just models or apps.
• "Engineer" means writing code, designing tools, and fixing bugs.

"Most healthcare tech is built by engineers who don’t understand clinics—or clinicians who don’t build. I do both."

I apply clinical problem-solving to technical systems:

1. Start with the problem, not the tool.
2. Diagnose bugs like a clinician: data, model, or deployment?
3. No black boxes—if we can’t explain it, it’s not ready.
4. Test safely: shadow deployments and A/B tests, like clinical trials.

I’m open to collaborating on:

• Open-source non-profit automation
• Local/offline tools for resource-limited settings
• Reproducible, ethical automation

Contact: [Email](k.jacoby at posteo.de)

Find my work:

• Kaggle
• Hugging Face

I contribute to open-source projects that make a real difference.

"The future of healthcare isn’t about replacing professionals with technology. It’s about giving them better tools—built together, for everyone."

---