Python GenAI Image Recognition Platform

I’m building a production-ready image-recognition web app that works exclusively with photographs. The core tasks are facial recognition and general image classification; object detection could be added later, but it is not part of this first milestone.

You’ll take the project end-to-end: model selection or fine-tuning (TensorFlow / PyTorch, OpenCV, or a comparable GenAI framework), a RESTful API layer, and a lightweight front-end where users can upload images and see results. I already have an AWS account plus a small labelled photo set; you’ll help expand the dataset and optimise accuracy.

The workflow I’m envisioning:
• Users upload a photo → backend securely stores it, runs preprocessing, then either returns a face-match report or a classification label set with confidence scores.
• All calls are authenticated with JWT and logged.
• Results must come back in under half a second for a 1080p image on a t3.medium instance.

Deliverables
• Clean, well-commented Python code (preferably FastAPI) packaged in Docker
• Front-end (React / Next.js or similar) for upload and result display
• Unit & integration tests covering 80 % of the codebase
• README and deployment script for AWS (CloudFormation or Terraform)

Acceptance criteria
1. ≥ 90 % accuracy on the validation set I’ll provide
2. Latency target met for both facial recognition and classification endpoints
3. CI pipeline passes on every push
4. Complete hand-off with documented setup, model re-training, and rollback steps

I’d like to see a working demo within two weeks and a full release six weeks from project start. Let me know which frameworks you prefer and any prior projects that show you can hit these performance numbers.