Applied AI engineer with foundations in data engineering and machine learning.
Portfolio · LinkedIn · All projects
I hold an M.Sc. in IT & Cognition from the University of Copenhagen. My work spans data ingestion and processing, feature engineering, model evaluation, and LLM applications built around retrieval and tools. I focus on making results traceable and testing where systems fail.
I'm eager to learn, quick to adapt, and resilient in competitive environments. I work well both independently and as part of a team.
Open to Applied AI, AI Solutions, and AI product engineering roles.
| Project | Engineering focus | Explore |
|---|---|---|
| RAGOps Lens | RAG retrieval, evaluation, confidence gating and observability with FastAPI, pgvector and Qdrant. | Architecture and evaluation |
| GitHub Opportunity Miner | LangGraph + FastAPI + Next.js agent turning GitHub evidence into source-linked opportunity cards and validation plans. | Workflow and demos |
| Fund Facts Cross-Check | Two-model comparison with structured outputs, citation checks and regression evals. Separates agreement from evidence support. | Video demo · Eval report |
| AML Transaction Review | XGBoost transaction ranking, temporal validation and review-budget evaluation; Databricks batch scoring with MLflow and Delta. | Experiment · Acceptance evidence |
| DeviceCare Decision Benchmark | Reproducible comparison of two models for device-support routing, with policy checks, repeated runs and documented review limits. | Report |
| M.Sc. Thesis: Retrieval and Agent Utility | Research on experience retrieval for an LLM agent, with frozen results, analysis and reproduction checks. | Results and verification |
These repositories include prototypes, benchmarks and research. AML, DeviceCare and Fund Facts use synthetic data; opportunity cards are hypotheses, not evidence of customer demand. Repository READMEs document evaluation boundaries and AI assistance where applicable.
- Data engineering: data ingestion and cleaning, SQL, batch pipelines, data validation and repeatable writes. ERP Risk MLOps and AML Transaction Review include Databricks / Delta batch workflows and MLflow model tracking.
- Machine learning: feature engineering, classification, anomaly detection, temporal validation and imbalanced-data evaluation, including experiments with rules, Isolation Forest and XGBoost.
- Research and deep learning: multimodal speech emotion recognition, multilingual text detoxification and gaze-informed retrieval.
- Data & ML: Python · SQL · scikit-learn · XGBoost · MLflow · Databricks / Delta
- AI applications: RAG · LangGraph · structured outputs · citation validation · regression evals
- Delivery: FastAPI · PostgreSQL · pgvector / Qdrant · Docker · Azure · TypeScript / Next.js