At ÄIO, I built and maintained Python machine-learning workflows for data preparation, model training, and evaluation. I used MLflow to track experiments and model versions, and added testing and evaluation checks to help catch regressions before deployment.
At Lingvist, I developed machine-learning and NLP workflows for language data, using embeddings and semantic search to retrieve relevant content. I integrated model inference into backend workflows and APIs so application components could use model outputs.
On the freelance Munichat project, I implemented the complete AI and backend functionality for a chatbot serving US townships. Residents could use the chat interface to get updated information drawn from township documents and website content.
