At Pro Codes Solutions, I design and deploy LLM-powered solutions for document handling, reporting, and customer communication, cutting manual effort by up to 60%. I build RAG pipelines that ground answers in internal documents and develop NLP extraction workflows that reach 90%+ accuracy on validation sets.
I also create evaluation harnesses to track accuracy, hallucinations, and regressions, and tune prompts, chunking, and model selection to lower inference costs by about 35%. I expose these workflows through FastAPI services containerized with Docker.
Through my self-employed AI/ML work, I deliver projects for international clients, from requirements and data analysis through deployment and handover. I’ve built assistants that resolve about 70% of routine queries and Python automations that save clients 10+ hours per week.

