Where developers grow together

MedRush

MedRush is a real-time emergency medical response and assistance platform designed to reduce critical response time during health emergencies. The app connects users to nearby hospitals, ambulances, blood donors, and emergency contacts instantly using GPS-based location intelligence. Core Purpose To bridge the gap between emergency situations and immediate medical support using smart technology. Key Features 🚑 One-tap Emergency SOS with live location sharing 🏥 Nearby Hospitals & Clinics Finder (GPS powered) 🩸 Blood Donor Search & Emergency Requests 📍 Real-time Ambulance Request System 👨‍⚕ Digital Medical Profile (store allergies, conditions, blood group) 🔔 Instant Alerts to Emergency Contacts 🔐 Secure and encrypted data handling Why MedRush? In medical emergencies, every second matters. MedRush is built to reduce panic, automate emergency communication, and improve survival probability through instant action.

NEXUS The Future of AI-Powered Talent Acquisition.

🏢 Nexus HR Intelligence: AI-Powered Resume Screener "Automating the first glance. Fair, Efficient, and Explainable Talent Acquisition." 🌟 The Vision In a world where one job posting attracts thousands of resumes, the traditional hiring funnel is broken. Recruiters spend an average of 6 seconds per resume, leading to burnout and missed opportunities. Nexus HR Intelligence is an enterprise-grade AI solution that reads, scores, and ranks candidates instantly—transforming weeks of work into seconds. What Makes Nexus Special? Semantic Search (No Keywords): We don't just count keywords. Our NLP model uses Sentence Transformers (Hugging Face) to understand the meaning of a candidate's experience, matching it to the Job Description with advanced semantic similarity. Explainable AI: Nexus doesn't just give a score; it generates a radar symmetry map for every candidate, showing exactly why the AI ranked them that way (e.g., strong experience, weak achievements). Secure Multi-Tenant Auth: Built on MongoDB Atlas, Nexus allows different companies (like Stark Industries or Wayne Enterprises) to register and manage their own isolated talent pools securely. 🛠️ Tech Stack & Architecture This is a Full-Stack Microservices Application built for high-performance AI inference. Frontend: Streamlit (UI, Visualizations, Auth UI, Glassmorphism CSS) Database: MongoDB Atlas (Cloud NoSQL DB for users and candidates) NLP & Scoring: Sentence Transformers (all-MiniLM-L6-v2) via Hugging Face Data Science: Scikit-learn (Linear Regression), Pandas, Numpy, Plotly Backend: FastAPI & Uvicorn (REST API) 🚀 How to Run Locally If you wish to run the development environment on your machine: Deployed Website Link: https://analyserresume.streamlit.app/ Clone the repository: git clone https://github.com/your-username/your-repo-name.git cd your-repo-name Install dependencies: We recommend using a virtual environment (venv). pip install -r requirements.txt Set up the environment variables: Create a .env file in the root directory and add: MONGO_URI=your_mongodb_atlas_connection_string Start the Backend: uvicorn api:app --reload Start the Frontend (New Terminal): streamlit run app.py 👥 The Team (SAH_Hackathon) Devkanti Sarkar- Resume Parsing and Extraction Engineer Mohar Gorai - Scoring Algorithm and Data Processing Engineer Agrima Gupta - System Integration and UI developer

SAH_SyntaxX

This project presents a deep learning–based off-road terrain segmentation and navigable path extraction system built using the DINOv2 Vision Transformer (ViT-B/14). The system performs pixel-wise semantic segmentation on off-road terrain images and intelligently extracts smooth navigable lane guidance suitable for autonomous vehicle applications. A pretrained DINOv2 backbone was leveraged to capture high-level visual representations of complex terrain patterns. A custom convolutional decoder was designed to generate dense segmentation maps for 10 terrain classes. To improve segmentation quality and handle class imbalance, a combined loss function consisting of Cross-Entropy, Dice Loss, and Focal Loss was implemented. The trained model achieved a Mean Intersection over Union (mIoU) of approximately 0.60 on the validation dataset. Post-processing techniques such as connected component analysis and polynomial curve fitting were applied to extract the largest drivable region and generate smooth left and right boundary curves along with a center navigation path.