Hate-Speech Detection Model & Setup

We are a high school team working on an AI capstone project and are looking to outsource the development of a hate-speech detection model. The project involves creating a text classification model that can accurately detect hate speech and includes two key features: - Feedback Mechanism: A simple system where users can flag false positives (non-hate speech classified as hate) and false negatives (missed hate speech). The flagged data should be stored for retraining the model. - Custom Data Training: The model should support retraining using our custom-labeled dataset to improve its accuracy over time. Deliverables: - A working hate-speech detection model (pre-trained or built from scratch using NLP models like BERT, RoBERTa, etc.). - Integration of a feedback mechanism that stores flagged data in a database (preferably in a simple format like SQLite). - Instructions and support for retraining the model with custom data. - Deployment support (e.g., on Heroku or AWS). Skills Required: - Experience in NLP (Natural Language Processing) and text classification models. - Familiarity with transfer learning (preferred frameworks: Hugging Face, TensorFlow, or PyTorch). - Experience in building or fine-tuning models using BERT, RoBERTa, GPT. - Ability to integrate backend (Flask, Node.js) for the feedback mechanism and store user feedback. - Knowledge of model deployment on cloud platforms (e.g., AWS, Heroku).