Led a team of around 7 that developed an automated system to use optical character recognition (OCR) to ingest PDFs or images of submitted government documents.
I started the project as a developer. I wrote a machine learning model to classify different form types that we supported and most of the backend pipeline to SQL. As the project got bigger, I was given a leading role and more developers were brought on. As tech lead I met with regularly with stakeholders to share progress, write up requirements for new features, worked with my team to design architecture proposals, and helped with development. The team trained more specialized machine learning models, designed a UI for government employees to verify and fix OCR output, and designed pipelines to recognize, apply schemas and ingest.
By the time I left the project we had saved tens of thousands of man hours of entering this information manually, according to feedback and estimates from our clients.Led a team of around 7 that developed an automated system to use optical character recognition (OCR) to ingest PDFs or images of submitted government documents.
I started the project as a developer. I wrote a machine learning model to classify different form types that we supported and most of the backend pipeline to SQL. As the project got bigger, I was given a leading role and more developers were brought on. As tech lead I met with regularly with stakeholders to share progress, write up requirements for new features, worked with my team to design architecture proposals, and helped with development. The team trained more specialized machine learning models, designed a UI for government employees to verify and fix OCR output, and designed pipelines to recognize, apply schemas and ingest.
By the time I left the project we had saved tens of thousands of man hours of entering this information manually, according to feedback and estimates from our clients.WWWWWWWWLed a team of around 7 that developed an automated system to use optical character recognition (OCR) to ingest PDFs or images of submitted government documents.
I started the project as a developer. I wrote a machine learning model to classify different form types that we supported and most of the backend pipeline to SQL. As the project got bigger, I was given a leading role and more developers were brought on. As tech lead I met with regularly with stakeholders to share progress, write up requirements for new features, worked with my team to design architecture proposals, and helped with development. The team trained more specialized machine learning models, designed a UI for government employees to verify and fix OCR output, and designed pipelines to recognize, apply schemas and ingest.
By the time I left the project we had saved tens of thousands of man hours of entering this information manually, according to feedback and estimates from our clients.
