 The study aims to develop an AI pipeline on digital images at the service to accurately diagnose treatable precancerous lesions in low- and middle-income countries, LMIC, through visual triage of women testing positive for human papillomavirus, HPV. The study implemented a comprehensive deep learning model selection, an optimization study on a large, collated, multi-geography, multi-institution, and multi-device dataset of 9,462 women, 17,013 images, and achieved an area under the receiver operating characteristics, ROC, CERV, AUC, of 0.89 within the study population of interest, with a limited total extreme misclassification rate of 3.4% on held-aside test sets. The model also produced reliable and consistent predictions, achieving a strong quadratic weighted kappa, QWK, of 0.86 and a minimal %2 class disagreement, %2 CLD, of 0.69% between image pairs across women.