Selected work
Each project shows what was built, the problem it solves and how it was taken from idea to a working application.
Problem — A multi-step process handled manually, with information spread across several tools and data sources.
Solution — An agent that plans the task, calls the right tools, reads external data and completes the workflow end to end, with a human checkpoint where it matters.
Python · LLM APIs · Agent framework · FastAPI
Problem — Knowledge locked inside documents that are slow to search and hard to query in natural language.
Solution — Documents are chunked and embedded; questions retrieve the most relevant passages, and the answer is generated with references to the sources.
Python · Embeddings · Vector database · LLM APIs · Streamlit
Problem — A decision that depended on intuition and spreadsheets, with historical data available but unused.
Solution — A supervised model trained on historical data, wrapped in an application that non-technical users can run and understand.
Python · Pandas · scikit-learn · Streamlit · Azure ML
Private project — production AI solutions developed for clients and companies are not public. Details available upon request.
Contact
Let's discuss it.