I summarized how I approached and reflected on processing a complex pharmaceutical vendor file during the externship.

I built Python notebooks in Google Colab to practice data structures, control flow, functions, data cleaning, JSON handling, text standardization, and image preprocessing for OCR.

I extracted text and structured fields from multi-page SDF PDFs using PyMuPDF, regex, and bounding-box grouping.

I compared Tesseract, PaddleOCR, and EasyOCR on a noisy scanned pharmaceutical PDF and reported which engine produced the most reliable field extraction.

I built and tested a RAG pipeline for document retrieval using LlamaIndex, implementing semantic chunking, embeddings, hybrid retrieval, and reranking.

I ran three RAG configurations on a sample SDF document and compared retrieval quality, filtering, and reranking to improve answer relevance.

I built page-level metadata extraction and document boundary detection for messy bundled PDFs, then refined prompts and rules to improve classification consistency.

I built a Gradio chatbot UI and connected it to a RAG backend for PDF question answering.

I built a Prompt Node that expanded short style ideas into structured image prompts for a moodboard generator agent.

I delivered a complete n8n workflow for a Consumer Trend Discovery AI Agent focused on Hallway Runner rugs, including input parsing, validators, data-fetching nodes, AI processing, and report packaging.

I delivered a Competitor Monitoring AI Agent that assembled retailer product data and produced a downloadable competitor intelligence report.

I updated the AI Content Agent by planning and specifying concrete prompt and form changes to make outputs sound like Wayfair's warm, lived-in voice.