My specialty is Natural Language Processing (NLP) using cutting-edge AI models like Large Language Models (LLMs), including Llama2, Mistral, and BERT. I am experienced in developing & deploying solutions utilizing RAG, using vector databases (Qdrant, Chroma db, PostgresSQL, etc) to unlock deeper meaning from text data via leveraging sentence transformers.
In one of my recent projects, I utilized llama2, Mistral LLM to extract information from pdf documents, developing specific prompts based on the nature of the information being extracted. Also, utilization of NER models that are trained for task-specific provides more accurate results but has an overhead of finetuning. A mix of both techniques can also be utilized to deliver results.
Questions:
Q1: are you open to hosting LLM’s locally if needed or OpenAI is the only option?
Q2: what are the source document formats as data curation is a key success factor, good quality curation provides better results?
Q3: Provide details of DBMS needed and does chosen RDBMS has vector storage capabilities.
I have experience with python, flask, uvicorn and fastapi libraries, Golang and C#.