Casablanca-Settat, Maroc
Page
1
of
3
My internship was about trying to build a web app that would allow a client
to feed it some data (tables or textual), and ask questions to an AI agent
regarding that data. My work focused on The Retrieval-Augmented-Generation
(RAG) pipeline, which is the key of the process, as it enables the agent to
work on data it hasn't been trained on, thanks to the possibility of feeding him
additional context (i.e, some sort of data). I also worked on the vector search,
so that the AI may select what is relevant in the document it receives. Another
key component of this work is enabling the Ai to perform SQL methods on the
document (in relevant cases), so that it could answer more specific questions.
This step sparked multiple safety questions regarding the access of data, but
also the commands that the agent should or should not be able to do. I found
this step particularly engaging and stimulating. Lastly, I had the opportunity
to work on making sure that the agent's answers were user-friendly, and
formatted in a way that made them understandable even outside of our
technical field.
I had a glimpse of the web app and its UI, but unfortunately, there was not