PDF Parsing Expert Needed
Budget: €1,500 – €3,000 EUR
I am looking for a PDF parsing expert.
IDEAL SKILLSET
- Python (vectors + fuzzy match experience)
- PostgreSQL (to store data)
- API (feed data from database to UI + feed data from third party sources to DB + allow customers to feed data)
- Front end (idealy FastAPI)
PROJECT
In a nutshell, we are to parse few standard type digital PDF forms for standard datapoints
Forms are standard but the content in the forms varies and not all forms contain all datapoints.
Data is then stored in DB and fed either to user profile in UI or feed via API is made available to a selected range of data.
All datapoints can be linked to section name. Section name indicates anchor words that will be in the file. Anchor words suggest what datapoints needs to be extracted. Hence, we can employ vectors to find information. Fuzzy match is needed to deal with variations.
Simple explanation would be. If each file is a basket, basket can have apples (one level data = apples), each basket can have multi colour apples (two level data = apples + colour of each apple), each apple can have from 1 to 50 seeds (three level data = apples + colour of each apple + seed count in each apple).
Objective is to show most recent datapoint for all PDF files parsed (if seed count changed for red apple, we need to show the latest seed count)
CHALLENGES
Current code to parse data needs refactoring and/or re-writting. Engine to pull data is 95% complete. There are few challenges that could be resolved with fuzzy match.
RESOURCES
Current code can be used to explain how data is to be extracted and how it is to be displayed.
IDEAL SKILLSET
- Python (vectors + fuzzy match experience)
- PostgreSQL (to store data)
- API (feed data from database to UI + feed data from third party sources to DB + allow customers to feed data)
- Front end (idealy FastAPI)
PROJECT
In a nutshell, we are to parse few standard type digital PDF forms for standard datapoints
Forms are standard but the content in the forms varies and not all forms contain all datapoints.
Data is then stored in DB and fed either to user profile in UI or feed via API is made available to a selected range of data.
All datapoints can be linked to section name. Section name indicates anchor words that will be in the file. Anchor words suggest what datapoints needs to be extracted. Hence, we can employ vectors to find information. Fuzzy match is needed to deal with variations.
Simple explanation would be. If each file is a basket, basket can have apples (one level data = apples), each basket can have multi colour apples (two level data = apples + colour of each apple), each apple can have from 1 to 50 seeds (three level data = apples + colour of each apple + seed count in each apple).
Objective is to show most recent datapoint for all PDF files parsed (if seed count changed for red apple, we need to show the latest seed count)
CHALLENGES
Current code to parse data needs refactoring and/or re-writting. Engine to pull data is 95% complete. There are few challenges that could be resolved with fuzzy match.
RESOURCES
Current code can be used to explain how data is to be extracted and how it is to be displayed.