Profile
LLM Engineer and BI Developer with 5 years building production AI and data systems in the Microsoft Azure stack inside a German healthcare company. Built and deployed an Azure Functions application using GPT-4.1 and LangChain that automates document classification, indexing, and case summarisation against an Enaio DMS — processing approximately 100 medical documents per day at 85% accuracy, reducing per-document handling time from 30 minutes to seconds. Designed and maintain 20+ Power BI dashboards across 8 departments integrating Navision ERP, Enaio DMS, and Microsoft HR. Author on five peer-reviewed healthcare publications (1 published, 4 under review), comfortable bridging clinical domain requirements and data infrastructure. Open to AI Engineering, Data Science, and ML Engineering roles in Germany, the Netherlands, and Switzerland.
Work Experience
reha assist GmbH · Arnsberg, NRW, Germany
LLM & AI Development (2024 – Present)
- Built Azure Function App (GPT-4.1, LangChain, Azure OpenAI Service) to automate Enaio DMS document workflows: classification, indexing, and case summary extraction — approximately 100 documents per day at 85% accuracy, reducing handling time from 30 minutes to under 5 seconds per document.
- Applied deliberate model selection strategy: GPT-4.1-nano for OCR pre-processing (cost efficiency), GPT-4.1 for complex multi-step reasoning on case summaries and Aktenspiegel generation — reducing API cost while maintaining accuracy.
- Currently developing document privacy filter using open-source LLM model to redact sensitive patient data before cloud LLM processing; privacy-by-design in regulated healthcare AI pipelines.
- Part of a 4-person development team; responsible for LangChain integration, Azure Functions deployment, and prompt engineering architecture.
Business Intelligence (2021 – Present)
- Designed and maintain 20+ Power BI dashboards across 8 departments (revenue, employee performance, absence management, sales analytics) used by C-level for monthly business reviews.
- Integrated Microsoft Navision ERP, Enaio DMS, and Microsoft HR into a unified cross-departmental reporting layer using DAX models and ETL pipelines.
- Automated bonus calculation process, reducing manual calculation from 3 days to an instant automated view.
Healthcare Data Science (2022 – Present)
- Corresponding author on peer-reviewed paper submitted to Das Gesundheitswesen (2026): effectiveness of telephone-based rehabilitation follow-up across 408 patients; functional limitation score reduced 56% (Cohen's d = 1.37; p < 0.001).
- Co-author on 4 additional healthcare publications (1 published in Die Unfallchirurgie Oct 2025; 3 currently under review in BMC Health Services Research, Die Unfallchirurgie, and Zeitschrift für Gerontologie und Geriatrie).
- Built ML classification model predicting Nachbehandlungsempfehlungen from ICUC pain and functional limitation scores using Python and scikit-learn.
- Bridge between clinical staff (doctors, department heads) and data infrastructure — translating domain requirements before building analytical pipelines.
reha assist GmbH · Arnsberg, NRW, Germany
- Enaio DMS configuration and ETL workflow design; foundational Power BI reporting and data modelling.
Tehran Municipality · Tehran, Iran
- Helpdesk lead for 1,000+ employees; achieved 15% improvement in system uptime.
- Smart City project (District 6): delivered 10% utility cost reduction; built SQL and Elasticsearch dashboards reducing reporting time by 20%.
Personal Projects
- Autonomous 24/7 platform for BTC/ETH options on Deribit: ML-based market regime detection (LightGBM), displacement scanner, on-chain analytics (GEX/DEX, max pain, IV surface), 4 strategy evaluators, Telegram real-time alerts.
- Deployed on Hetzner VPS with systemd process management and PostgreSQL data collection daemon storing options trades every 30 minutes.
- Stack: Python 3.13, LightGBM, XGBoost, PostgreSQL, PySide6.
- GitHub: github.com/nick9248/rsc_option_trading
- Python bot mirroring trades from target wallets on Polymarket prediction market CLOB using EIP-712 signed blockchain transactions (POLY_PROXY signature mode).
- Multi-layer safety system: geo-verification, 10% slippage guard, near-expiry filter, genesis block protection, on-chain execution validator.
- Stack: Python, Polymarket CLOB API, Web3/EIP-712, PostgreSQL, VPS deployment.
- GitHub: github.com/nick9248/polymarket_bot
Publications
Long-term care of polytraumatized patients, requirements for rehabilitation management
Die Unfallchirurgie · October 2025 · Co-Author · Published
Detecting disorders early in the healing process: Telephone support during the healing process using the ICUC score
BMC Health Services Research · Co-Author · Under peer review
Effectiveness of Telephone-Based Health Visits in Musculoskeletal Rehabilitation
Das Gesundheitswesen · Corresponding Author · Under peer review
Langzeitbetreuung polytraumatisierter Patienten — Teil 2
Die Unfallchirurgie · Co-Author · Under peer review
Pflegeberatung – Stärken und Schwächen
Zeitschrift für Gerontologie und Geriatrie · Co-Author · Under peer review
Education
Bergische Universität Wuppertal · Wuppertal, Germany
Focus: Data Analysis, Mathematics, Financial Mathematics · Pursued alongside full-time employment · 3 courses remaining
Kharazmi University · Tehran, Iran
Focus: Machine Learning, Data Analysis, Mathematics
Skills
LLM / AI: GPT-4.1, GPT-4.1-nano, LangChain, Azure OpenAI Service, RAG, Prompt Engineering, NLP, Document AI
Machine Learning: LightGBM, XGBoost, scikit-learn, logistic regression, statistical analysis
Business Intelligence: Microsoft Power BI, DAX, Power Query, ETL, data modelling, Navision ERP
Cloud / Infra: Azure Functions, Azure OpenAI, Hetzner VPS, systemd, Nginx
Programming: Python 3.13, SQL (CTEs, joins), PostgreSQL, Git, REST APIs
Other: Enaio DMS, Web3/EIP-712, Telegram Bot API, PySide6
Languages
- English — Professional working proficiency
- German — B1 TELC certificate (actively advancing to B2)
- Persian (Farsi) — Native
Certifications
- Claude Code in Action — Anthropic (February 2026)
- Claude Code: Software Engineering with GenAI Agents — Vanderbilt University (October 2025)
- Generative AI Automation Specialization — Vanderbilt University (December 2024)
- Agentic AI and AI Agents: A Primer for Leaders — Vanderbilt University (February 2025)
- Deploy and Maintain Power BI Assets — Microsoft (February 2025)
- Python for Data Science, AI and Development — IBM (May 2024)
- Azure AI Engineer Associate (AI-102) — In progress, target Q3 2026