I built an automated OCR + CV + OpenAI pipeline that replaced ~10 months of manual extraction:

  • Document cleanup & segmentation
  • OCR pass + structure mapping
  • LLM-assisted validation

Result: delivered ahead of schedule; increased research throughput substantially.