requirements.txt

# [ML Core]
torchvision==0.17.0  # Pairs exactly with PyTorch 2.2
transformers==4.38.2  # Stable version for DINOv2 and BioBERT
datasets==2.18.0  # Required for NCBI dataset parsing
accelerate==0.27.2  # Handles memory-efficient model loading
peft==0.9.0  # Needed for lightweight fine-tuning parameter handling
sentence-transformers==2.6.0  # powers the all-MiniLM-L6-v2 embedding model
timm==0.9.16  # Required backend for DINOv2 architecture
scikit-learn==1.4.1.post1  # Used for metric calculations (F1, precision)
numpy==1.26.4  # Last stable 1.x version before 2.0 breaking changes
Pillow==10.2.0  # Safe version for PIL image processing
opencv-python-headless==4.9.0.80  # Headless avoids Qt DLL conflicts on Windows servers
openai-whisper==20231117  # Reliable version for Whisper CPU transcription

# [Backend]
fastapi==0.110.0  # Base API framework
uvicorn[standard]==0.27.1  # ASGI server for FastAPI
python-multipart==0.0.9  # Required for FastAPI file uploads
pydantic==2.6.3  # Pydantic v2 exact version for type safety
pydantic-settings==2.2.1  # Required for v2 configuration management
slowapi==0.1.9  # Rate limiting without Redis
python-jose[cryptography]==3.3.0  # JWT creation and verification
passlib[bcrypt]==1.7.4  # Password hashing algorithms
httpx==0.27.0  # Async HTTP client for OAuth
aiofiles==23.2.1  # Async file handling for image uploads
authlib==1.3.0  # Handles Google OAuth 2.0 flow
itsdangerous==2.1.2  # Required by Authlib sessions
python-magic-bin==0.4.14  # Windows-compatible MIME type checker (critical)

# [Database]
sqlalchemy==2.0.28  # SQLAlchemy 2.0 for async ORM
alembic==1.13.1  # Database migration manager
aiosqlite==0.20.0  # Async SQLite driver for local dev
asyncpg==0.29.0  # Async PostgreSQL driver for production
greenlet==3.0.3  # Required for SQLAlchemy async execution

# [RAG + Vector DB]
chromadb==0.4.24  # Stable local vector database

# [Utilities]
reportlab==4.1.0  # PDF report generation
fpdf2==2.7.9  # Fallback lightweight PDF generation
python-dotenv==1.0.1  # Environment variable parsing
tqdm==4.66.2  # Progress bars for dataset processing
ffmpeg-python==0.2.0  # Whisper audio conversion wrapper
apscheduler==3.10.4  # In-memory background task scheduling

# [Dev + Testing]
pytest==8.1.0  # Core testing framework
pytest-asyncio==0.23.5  # Allows testing async FastAPI routes
black==24.3.0  # Code formatting
ruff==0.3.4  # Fast Python linter
mypy==1.9.0  # Static type checking
locust==2.24.0  # Load testing framework