I explained how Large Language Models work, summarizing their training, architecture, and post-training refinements.

I prepared and cleaned pharmaceutical document data using Python, applying Pandas and image preprocessing techniques to improve OCR readiness.

I extracted and structured key fields from multi-page SDF PDFs using PyMuPDF, producing a repeatable data pipeline.

I converted scanned pharmaceutical PDFs into structured, validated text and rebuilt fields from noisy OCR output.

I built and tuned a Retrieval-Augmented Generation pipeline using LlamaIndex, implementing chunking, embeddings, query processing, and retrieval optimization.

I compared embedding models and RAG configurations, then evaluated open-source LLMs in my pipeline.

I built and stabilized a PDF segmentation and routing pipeline that split bundled regulatory PDFs into page-level documents, classified pages with doc_type tags, and routed queries to the correct documents.

I built a Gradio chatbot interface and connected it to a retrieval-augmented pipeline that answered questions from a PDF.