Implemented Backend API

[blueee04]

Implemented all required backend API routes, Pydantic schemas, CrewAI agent pipeline (Researcher/Analyst), Mock Expert Matching, and Supabase database client.

[Louis047]

Hey there, the PR has a solid base and almost everything is integrated well but from my testing, I have found some minor issues listed below:

• Mocked Academic Search: The AcademicSearchTool in tools.py uses a hardcoded time.sleep(1) mock response. This is perfectly fine for now as an MVP step, but we will have to to wire it up to the real Semantic Scholar API. This is purely kept for notice and would be fine for an MVP as said earlier.

• Local Windows Testing Issues: I attempted to run pip install -r requirements.txt to spin it up locally and test the endpoints, but it failed to compile tiktoken and regex wheels on this Windows machine (this is a common issue with heavy AI libraries like CrewAI on Windows).
  Recommendation: Since you have a Dockerfile setup, it's heavily recommended that you run the backend using Docker (e.g. docker-compose up backend) to completely bypass these Windows-specific Python C++ issues.

• Requirement Issue: In requirements.txt, I noticed an invalid crewai>=0.86.0 version constraint (which caused pip to fail initially). I had to remove that constraint to proceed. Either drop the version constraint or pin it to a stable one like crewai>=0.80.0.

Let me know if I have missed anything else, I just did some basic testing for now and the above are my notes on what I have noticed 👍

[blueee04]

Alr check this out