Migrate Backend to JetPack 6 with Python 3.10

[rajjainl]

This pull request was generated by @kiro-agent 👻

Comment with /kiro fix to address specific feedback or /kiro all to address everything.
_{Learn about Kiro autonomous agent}

---

Summary

This PR updates the backend for JetPack 6 Orin compatibility, migrating from Python 3.9 to Python 3.10 (the native interpreter for JetPack 6).

JetPack 6 Specifications

Component	Version
OS	Ubuntu 22.04
Python	3.10 (native)
CUDA	12.6
TensorRT	10.3
cuDNN	9.3

Why Python 3.10 (not 3.11)?

1. JetPack 6 ships with Python 3.10 as the default interpreter on Ubuntu 22.04
2. NVIDIA's pre-built wheels (PyTorch, TensorRT bindings, etc.) for JetPack 6 are compiled for Python 3.10
3. Using Python 3.11 would require building NVIDIA libraries from source, causing compatibility issues

Changes

Dockerfile (src/backend/Dockerfile)

• Updated base image to Ubuntu 22.04 (JetPack 6 native OS)
• Changed from Python 3.9 (compiled from source) to Python 3.10 (apt package)
• Improved Dockerfile structure with better layer caching
• Added proper pip requirements installation
• Removed legacy libssl1.1 dependency (not available in Ubuntu 22.04)
• Removed OS 18.04 specific code
• Updated CMD to use generic python3 instead of python3.9

requirements.txt (src/backend/requirements.txt)

Package	Old Version	New Version	Notes
Jetson.GPIO	2.0.21	2.1.0	JetPack 6 compatible
awsiotsdk	1.11.9	1.21.0	Latest stable
marshmallow	3.19.0	3.21.0
pyyaml	5.4.1	6.0.1	Python 3.10 compatible
grpcio	1.56.2	1.62.0	Aligned versions
grpcio-tools	1.51.1	1.62.0	Aligned with grpcio
jsonschema	4.17.3	4.21.0
numpy	1.24.3	1.26.4	Python 3.10 optimized
pycairo	1.23.0	1.26.0
PyGObject	3.42.2	3.48.0
psutil	5.9.5	5.9.8
fastapi	0.109.2	0.110.0
uvicorn	0.23.2	0.28.0
pydantic	2.1.1	2.6.0
asgi-correlation-id	4.2.0	4.3.0
structlog	22.1.0	24.1.0
sqlalchemy	2.0.21	2.0.28
alembic	1.12.1	1.13.1
httpx	0.24.1	0.27.0
scikit-learn	1.0.2	1.4.0	Python 3.10 support

Important Notes

DLR (Neo Deep Learning Runtime)

• dlr==1.10.0 is retained but may need custom build for JetPack 6
• Check neo-ai-dlr for JetPack 6 wheels

Jetson.GPIO

• Upgraded to 2.1.0 for JetPack 6 compatibility
• Should work with Orin GPIO configuration

Testing Recommendations

1. Build Docker image on JetPack 6 Orin device:

   docker build -t dda-backend:jetpack6 --build-arg OS=22.04 -f src/backend/Dockerfile src/backend/

2. Verify Python version inside container:

   docker run --rm dda-backend:jetpack6 python3 --version
   # Expected: Python 3.10.x

3. Test NVIDIA library compatibility (TensorRT, CUDA) on the Orin device