SHREYA AGRAWAL

1.Mental Fatigue Detector | AI & Full-Stack Web App

Python • Django • Scikit-learn • OpenCV • JavaScript

• Problem solved: Mental fatigue goes undetected until it impacts productivity — built a multi-modal AI system that detects fatigue in real time by analyzing typing patterns, mouse movements, and facial expressions simultaneously.

• ML Core: Designed a custom ensemble ML pipeline combining behavioral signals from 3 data sources; implemented confidence scoring and z-score normalization for reliable fatigue prediction with personalized recommendations.

• DevOps pipeline: Built RESTful API backend in Django serving ML inference results to a dynamic dashboard; integrated WebRTC for real-time webcam-based blink rate and eye closure detection — all processed locally for privacy.

• Agentic pipeline design: Architected a multi-source signal orchestration system where independent behavioral agents (typing, mouse, facial) each process input streams and pass confidence scores to a central aggregator — mirroring Human-in-the-Loop agentic workflow design.

2.Full-Stack Food Delivery Platform | End-to-End Web Application

Python • Flask • JavaScript • MySQL • REST APIs • HTML/CSS

• Problem solved: Built a production-grade multi-role food ordering ecosystem — customers browse menus and place orders, restaurants manage inventory and accept/reject orders, delivery agents track assignments, and admins oversee the entire platform from a dedicated panel.

• Engineering depth: Architected RESTful APIs with clean separation between front-end and back-end; implemented role-based access control (RBAC) with secure session management ensuring each user type sees only their authorized views and data.

• Full feature scope: Includes real-time order status tracking, dynamic cart with price calculation, payment flow integration, order history, restaurant dashboard with live order queue, and an admin panel for user and content management — 6 modules, 3 user roles.

• API integration layer: Designed a modular REST API backend with clean interface contracts between frontend and backend — architecture compatible with third-party platform integrations (payment gateways, notification services).

3.Smart Sales & Customer Insight System | ML-Powered Analytics Platform

Python • Pandas • Scikit-learn • SQL • Matplotlib/Seaborn • Flask

• Problem solved: Businesses struggle to extract actionable intelligence from raw sales data — built an end-to-end analytics platform that ingests sales records, runs ML-based customer segmentation, and surfaces insights through an interactive dashboard.

• Engineering depth: Designed scalable ETL pipelines in Python/Pandas for data ingestion, cleaning, and feature engineering; applied clustering algorithms (K-Means) for customer segmentation and regression models for sales forecasting, with results visualized in a Flask-served dashboard.

4.Hospital Management System | DevOps & Full-Stack

Java • Jenkins • Docker • CI/CD • MySQL • REST APIs • OOP Design Patterns

• Problem solved: Manual hospital workflows are error-prone and slow — built a fully automated system handling patient registration, doctor scheduling, appointment booking, billing, and medical record management in a single unified platform.

• Engineering depth: Designed using OOP principles and layered architecture (Controller → Service → Repository) in Java, with a relational MySQL schema managing patients, doctors, appointments, and billing with referential integrity and optimized queries.

• DevOps pipeline: Configured a Jenkins CI/CD pipeline integrated with Git — automating code builds, unit test execution, and Docker container deployments on every commit, reducing manual deployment effort and eliminating environment inconsistency.

• Introduction to Internet of Things - NPTEL, IIT Kharagpur
• IBM Hackathon
• NodeRED - Udemy
• Frontend Skills - Infosys
• Fundamentals of MongoDB - MongoDB