I explained how large language models work, covering tokenization, embeddings, self-attention, the Transformer architecture, and pre-training versus fine-tuning.

I prepared and cleaned pharmaceutical-style data and images using Python in Google Colab, applying Pandas cleaning, JSON handling, text standardization, and image preprocessing for OCR readiness.

I extracted text and structured data from multi-page SDF PDFs using PyMuPDF and saved my outputs for review.

I compared Tesseract, PaddleOCR, and EasyOCR on a scanned pharmaceutical supplier PDF and reported which engine performed best for key-term extraction.

I built and tested a complete RAG pipeline using LlamaIndex, selecting components and running queries to validate retrieval.

I tested three RAG configurations on an SDF document and compared retrieval settings to find which returned the correct chunk.

I built page-level metadata extraction and document-boundary detection for bundled pharmaceutical PDFs.

I built a Gradio chatbot UI and connected it to a local RAG pipeline, adding a user-selectable background color and chat screenshots to document the interface.