You have unread messages about the job Virtual AI Clinician Diagnosis System

Hello and thanks for reaching out... the process of choosing the right fit for the project has been hard as alot of people claiming to know more than they actually do which made me do an assessment to understand the capability.

I am not fishing for free information, I am a Principal AI Engineer at a FAANG company and I can do the project end-to-end, I just dont have capacity to take it on. I a willing to pay the minimum of $100 or rate of 5 hours worth of work for successful submissions that adhere to details below.

I have completed that task in less than 3 hours, so it's fair task to assess knowledge before engaging in this project and other 5 projects needs to be worked on as well as filter people who know vs who claim to know

Objective: Develop a basic Retrieval Augmented Generation (RAG) system using an API (e.g. Google's Gemini, Open AI’s GPT..) to answer questions from a set of Arabic documents. The primary focus is improving the accuracy of answers in Arabic. You are free to do whatever you want to achieve that objective, which is improving the accuracy of the system’s answer for the questions it receives from users.

Submission Requirements
System Code and Documentation: Submit a link to the repository containing all the code used in the project within 3 days from receiving the test description. The repository should include a README file explaining the system architecture, data flow, and instructions for setting up and running the system. Please have the system exposed via a basic interface that we can test from, hosted on a free service.
Performance Report: Provide a report that includes the methodology used, experiments conducted, results obtained, and an assessment of the system's performance against the established metrics.

You can use any open source implementation for the interface. I used streamlit (already free examples that you can use to your own use case)

If interested please let me know and I will create a contract based on this test to see if we can move forward . Let me know if you have any questions (data for RAG is attached)