At The Stakeholder Company, I build and maintain Airflow data pipelines and reusable Python wrappers for external data provider APIs. I also built an end-to-end MLOps pipeline for a production model and a CircleCI deployment pipeline with staged promotion, approval gates, and rollback support.
I developed an entity resolution pipeline that matches identifiers deterministically when possible and otherwise uses candidate retrieval and an LLM for selection. It returns structured results with safeguards against hallucinated and low-confidence matches, achieving 93.23% accuracy.
At Uppsala University, I built data processing pipelines that turned UN Climate Change Conference materials into a text corpus. This included OCR for more than 2,000 PDFs, transcription for more than 100 recordings, and machine translation using external APIs.
At Holistics Software, I developed ETL/ELT pipelines into BigQuery and designed Snowplow event tracking for product analytics. I built SQL dashboards and led a team serving data requests; earlier, my work included data analysis across fintech, e-commerce, and market research.
