About

Welcome to the carecodeconnect home page! 🤗

“Building persuasive technologies with AI and Machine Learning for health, well-being & sustainability”

We combine the tools of social science and data science to build solutions to complex social problems.

  1. Decode the matrix of human relations with technology: by linking computation with interpretation, we tell the human stories behind the data.

  2. Build machine learning solutions to create caring futures: a data revolution that is equitable, fair, and sustainable. Our solutions are infused with care, ethics, privacy. We are sensitive to cultural, geographical, historical circumstances, social contexts and differences that make a difference.

  3. Apply cutting-edge free and open source technologies to put people at the heart of the data: creating evidence-based decision-making and actionable ‘outsights’ to make the world a better place.

In short, we care, code, connect.

Explore our blog and projects to find out more!

Recent Work Experience

  • Lead Data Scientist, Shoutt International, London (2026). Production LLM and agentic systems for a two-sided freelance job marketplace: role classification for thousands of inconsistent job posts, a CV-review engine with a multi-model router, an agent pipeline that turns one API call into a branded interview-prep PDF, and the evaluation harnesses and metrics dashboards that showed whether any of it worked.

  • Founding Developer, Life Repair Kit, Berlin / remote (2026–present). Sole developer of a German-language wellbeing platform: a single Rust/Axum binary serving a TypeScript SPA and PostgreSQL, the twelve-dimension ResilienzRadar self-assessment with on-device PDF results, a founder-editable resource hub and magazine, and the EU-only Scaleway infrastructure, CI/CD and security hardening behind it. System design and architecture write-up.

  • Data Scientist, Lightfoot, Exeter (2024–2025). Automated analytics, causal inference and predictive models for eco-driving behaviour change.

  • Data Scientist, Data Science Retreat, Berlin (2023–2024). Led a team of four to build Jhāna.AI, an interactive meditation assistant guided by real-time voice and brain signals; prompt engineering and fine-tuning of open-source LLMs.

  • Data Scientist, Clinical Psychology, Freie Universität Berlin (2022–2023). Led a research team building predictive models, data pipelines and interactive dashboards in Python and R, delivering insights to a stakeholder group of 60+ organisations.

  • Data Consultant, Institute for Applied Sustainability Studies, Potsdam (2017–2021). Social network analysis and an interactive data-visualisation dashboard of the sustainability field, for a project funded by the German Federal Ministry of Education and Research.

Skills

  • Full-stack data science for production. Research, model development, deployment, evaluation and monitoring across the lifecycle of a data product; software-engineering practice (testing, version control, CI/CD) applied to ML and AI pipelines at scale.

  • LLM and agentic systems that can be trusted. Prompt and context engineering, structured outputs, multi-model routing, tool-using agents, RAG — and the part that makes them shippable: golden datasets, evaluation harnesses, regression tests, drift dashboards and cost/latency measurement. Prompt for recall, deterministic code for precision.

  • Hybrid pipelines. Classical ML and rule-based NLP alongside generative models, each used where it wins on accuracy, cost or determinism.

  • Causal inference and experimentation. Experimental and quasi-experimental design, A/B testing, forecasting, and twenty years of teaching research methods — quantitative, qualitative and mixed.

  • Persuasive technology for behaviour change in health, well-being and sustainability, informed by human-computer interaction research.

  • Communication and leadership. Translating technical work into decisions for non-technical stakeholders; mentoring; leading small cross-functional teams.

Tools

Languages

  • Python (primary), C# / .NET 8, TypeScript / JavaScript, SQL (PostgreSQL, T-SQL, Cosmos DB), R, Bash, Rust

LLM & AI engineering

  • Azure AI Foundry and OpenAI Service, Anthropic Claude (Claude Code, MCP servers), OpenAI API, LangChain, Hugging Face, Ollama; prompt engineering, fine-tuning, agentic workflows, evaluation harnesses

Machine learning & statistics

  • scikit-learn, XGBoost, statsmodels, prophet, PyTorch, TensorFlow, spaCy / NLTK; causal inference, experimental design, time-series forecasting

Data stack

  • pandas, polars, NumPy, PySpark, pydantic; PostgreSQL, Azure Cosmos DB

Cloud, MLOps & DevOps

  • Azure (Cosmos DB, AI Foundry, OpenAI Service, Container Apps, Blob Storage, Application Insights, Azure DevOps Pipelines), AWS (S3, Lambda, EC2); Docker (multi-arch), GitHub Actions, CI/CD with production gates; self-hosted services on Raspberry Pi with Tailscale

Visualisation & BI

  • plotnine, Plotly, Observable Plot, matplotlib, ggplot2; Power BI, Zoho Analytics, Streamlit

Writing & documentation

  • Quarto (reproducible reports, slides, this site), pandoc, LaTeX, Jupyter, Sphinx