todo.md

To do

1.image -> 2.ocr -> 3.structured output -> 4.Background Search agent -> 5.dynamic vectorstore -> 6.Retriever

Done:

1.image to ocr 2. ocr to structured output

features:

1. image / images -> OCRs -> structured outputs (at max 5 at once)
2. search via imformations .. multiple vendors for each one ..
3. long tructured prompt
4. RAG plus background search agent

Bonus:

1. Save the images too in metadata of the structured output for further call MM RAG and show
2. deploy , ci/cd , dockerise (((( IMPORTANT)))
3. linting , logging
4. private memory + login (nginx?)

RAG with

distributed fast inference sys design topics Auto scaling dockerised deployed on aws ec2 fastHTML front-end

evaluation pipeline important