Offene Stellen
We are looking for a tech visionary who is enthusiastic about the project.
We are currently building our custom AIs in the ChatGPT customGPT infrastructure.
Here is where we are with this: We have a functioning data pipeline tool that draws metadata and data from an academic citation database (Zotero) and processes unstructured academic texts (pdfs) for use in RAG applications. It includes OCR and chunking of data for better citation retrieval in RAG system. This is currently a desktop version. We have also developed some pseudocode for customGPT instruction. This all works very well. You can see the current version here: scholarminds.ai > Directory > enter password “Schol@rMinds”.
Here are some pain points we are experiencing (1) Creation of customGPT is not scalable because data has to be uploaded manually. (2) Creation and management of users in Directory is not scalable because we have to do it manually. (3) RAG system and citation output could be improved incrementally in conjunction with input from people in digital humanities.
We have some development done that could be helpful here. Part of data pipeline with API to Zotero is completed. We have a user management system built. We have contacts in digital humanities and library science who can help us understand existing practices for dealing with unstructured data like books and articles.
We are currently building our custom AIs in the ChatGPT customGPT infrastructure.
Here is where we are with this: We have a functioning data pipeline tool that draws metadata and data from an academic citation database (Zotero) and processes unstructured academic texts (pdfs) for use in RAG applications. It includes OCR and chunking of data for better citation retrieval in RAG system. This is currently a desktop version. We have also developed some pseudocode for customGPT instruction. This all works very well. You can see the current version here: scholarminds.ai > Directory > enter password “Schol@rMinds”.
Here are some pain points we are experiencing (1) Creation of customGPT is not scalable because data has to be uploaded manually. (2) Creation and management of users in Directory is not scalable because we have to do it manually. (3) RAG system and citation output could be improved incrementally in conjunction with input from people in digital humanities.
We have some development done that could be helpful here. Part of data pipeline with API to Zotero is completed. We have a user management system built. We have contacts in digital humanities and library science who can help us understand existing practices for dealing with unstructured data like books and articles.