Putting ML models into production for page analysis, text-line-extraction, object detection, and HOCR of medival manuscripts.
Here you can find a variety of tools used to annotate data for ML, format data for ML, and running models in a UI. All projects are workspaces for The Babel Public Library.
You can also check out my basic project portfolio website MumbotPorts
Some of my favorite repos are pinned below, including a dataset I scrapped and formatted to mirror MINST but using a collection of 9 characters in latin textura from medieval text (provided by paleographers). An annotator aimed at leveraging a paleographers approach to transcribing, compiling, and carefully considering language data found in manuscripts. Exporters and API that convert the object structures I regularly use into standardized ML formats or standardized historic library formats such as PAGE XML, COCO, MARC, or Dublin Core. Last but not least API that may or may not be available to preform ML enabled alterations on datasets via lambda functions and sagemaker endpoints. (sagemakers endpoints are off more often than not cause thats a whole bill)
React
Node.js
AWS
Python
GitHub
Docker
Historic HOCR ML Pipelines !
Game asset generation !
Making DevOps Cheaper !
I'm really interested in natural language coding, few-shot-learning on depriciated data, unstructured language analysis, and just having fun with tech.
he/him
I love to bike in NYC, 12mi a day baby!