I wrote a concise explanation of how large language models work, covering training, token generation, and fine-tuning with human feedback.

I prepared and cleaned pharmaceutical documents using Python in Google Colab, applying data-structure, control-flow, Pandas, JSON handling, text cleaning, and image preprocessing exercises.

I extracted text and field-level data from multi-page SDF PDFs using Python, then applied regex and layout heuristics to locate key dates and vendor fields.

I compared Tesseract, PaddleOCR, and EasyOCR on a scanned pharmaceutical supplier PDF and documented preprocessing, extraction, and layout-aware differences.

I built a complete RAG pipeline for document retrieval using LlamaIndex and tested it end-to-end in a Colab notebook.

I ran three RAG configurations, compared outputs, and logged differences to inform retrieval tuning.

I built a page-level segmentation and routing prototype for bundled PDFs, extracting metadata and preparing document chunks for downstream pipelines.

I built a Gradio chatbot interface and connected it to a RAG pipeline to answer questions about a PDF, splitting UI and prompt logic into modular functions.