AI Content Pipeline 🤖🚀 Automates news article management. It fetches news, uses AI to rewrite/translate content, then provides a dashboard for review and approval. A full-stack pipeline from data to publication. ✨

This project implements a multimodal machine learning approach to predict housing prices using both structured tabular data and house images.

Project Overview
The model combines:

CNN-based image feature extraction from house images
Neural network processing of tabular housing data
Feature fusion to combine both modalities
End-to-end training for optimal performance
Features
Convolutional Neural Network (CNN) for image feature extraction
Dense neural network for tabular data processing
Feature fusion mechanism
Comprehensive evaluation metrics (MAE, RMSE)
Data preprocessing and augmentation
Training visualization and logging
Dataset

News Topic Classifier Using BERT
A powerful news classification system that uses fine-tuned BERT (Bidirectional Encoder Representations from Transformers) to categorize news headlines into four main topics: World, Sports, Business, and Sci/Tech.

🎯 Project Overview
This project demonstrates how to:

Fine-tune a pre-trained BERT model on the AG News dataset
Preprocess and tokenize text data for transformer models
Evaluate model performance using accuracy and F1-score metrics
Deploy the model with both Streamlit and Gradio web interfaces
Create an interactive web application for real-time news classification
✨ Features
BERT Fine-tuning: Uses bert-base-uncased as the base model
Multi-class Classification: 4 news categories with confidence scores
Interactive Web UI: Both Streamlit and Gradio interfaces
Real-time Predictions: Instant classification of news headlines
Visual Analytics: Confidence charts and probability distributions
Model Persistence: Save and load trained models
Comprehensive Evaluation: Accuracy, F1-score, and detailed metrics

Overview
This project implements an automated, end-to-end system to rank candidate resumes based on their match percentage against a given job description (JD). The core mechanism uses a Greedy Keyword Matching Algorithm combined with n8n automation to handle input, processing, and output.

The system functions as a lightweight AI tool for recruitment, automating the initial, time-consuming screening process.

✨ Features and Core Logic
Core Functionality
Input Handling: The system accepts a job description and multiple candidate resumes (PDF/TXT)
Greedy Scoring: Resumes are scored based on the ratio of JD keywords found in the resume.
Ranking & Reporting: The system returns a ranked list and automatically stores results in a spreadsheet/database while emailing the Top 3 candidates and their match percentages to HR[cite: 20, 21].

Problem Statement
Managing patient appointments, information requests, and scheduling for medical clinics can be time-consuming and error-prone when handled manually. Patients often face delays, miscommunication, or incomplete information when trying to book, reschedule, or cancel appointments. There is a need for a friendly, automated solution that can handle these interactions efficiently and naturally.

Solution: Receptionist Agent
The Receptionist Agent is a virtual assistant. It automates appointment booking, information requests, and scheduling tasks through natural, conversational interactions. The agent leverages AI and integrations with Google Calendar, Google Sheets, and internet search to provide a seamless experience for both patients and clinic staff.

Key Features
Conversational AI: Engages users in a friendly, professional manner to understand their needs and guide them step by step.
Appointment Management: Books, checks availability, and cancels appointments using Google Calendar.
User Data Handling: Retrieves and stores patient information securely in Google Sheets.
Knowledge Base: Answers general queries about the clinic, doctor, and services.
Reminders & Feedback: Sends reminders and collects feedback via WhatsApp or SMS.
Internet Search: Provides general public information when needed.
Contextual Memory: Remembers user sessions for a personalized experience.
How It Works
User Interaction: Patients interact with the agent via supported channels (e.g., WhatsApp).
Intent Analysis: The agent analyzes the user’s message to determine their goal (e.g., book, cancel, or inquire).
Step-by-Step Guidance: The agent asks for missing information only when needed, checks availability, and confirms actions.
Integration: Uses Google Calendar for scheduling, Google Sheets for user data, and internet search for general queries.
Confirmation: Always confirms actions and suggests alternatives if something is unavailable.