Document Understanding
OCR, handwriting recognition, layout analysis, classification, extraction, and evaluation for records that are noisy, old, scanned, or inconsistent.
Machine Learning Engineer | Document Understanding | PhD
Taylor Neil Archibald, PhD
I build reliable machine learning systems for documents, handwriting, historical records, and other high-variance data that standard automation handles poorly.
Profile
Research depth with production instincts.
My background combines PhD research in document understanding, enterprise machine learning experience at Ancestry, and hands-on engineering across Python, PyTorch, data pipelines, and web systems. I am strongest where model behavior, data quality, and business constraints all matter.
Professional Focus
Where I can contribute quickly
Applied ML Engineering
Training workflows, inference services, model diagnostics, data curation, and pragmatic quality gates for systems that need to keep working.
Research Translation
Turning papers, prototypes, and ambiguous technical requirements into usable software without losing the assumptions that make the work valid.
Selected Work
Document AI, handwriting, and engineering projects
Robust Historical Document Processing
Dissertation work focused on invariance, variance, and robustness in historical document understanding.
Simple HWR
Offline handwriting recognition implementation using neural sequence modeling and CTC-style training.
Fine-Grained Form Classification
Research on combining semantic segmentation masks and embeddings for census-style form classification.
Consulting
Need practical AI systems help?
I also work with teams that need focused help integrating machine learning, LLM workflows, automation, and document intelligence into messy real-world operations.
View consulting page