suchitha

Full-Stack Developer

Amrita Vishwa Vidyapeetham Bengaluru Campusfull_time, internship

Open to roles

REST APIs

MemoryVerse AI '26

This project is an AI-powered Digital Identity System that transforms scattered academic and professional documents into an intelligent, searchable knowledge repository. After installing the project dependencies and configuring the required environment variables (including a valid Anthropic API key), reviewers can upload sample documents such as resumes, certificates, internship letters, project reports, and portfolio files. The system automatically extracts content, categorizes documents into meaningful sections, identifies relationships between skills, projects, certifications, internships, and achievements, and generates a visual digital journey timeline. Users can retrieve information through natural language queries such as "Show my AI projects," "Show all my certificates," or "Show my latest resume," while preserving the original files in their uploaded format. The application uses semantic search and a knowledge graph to understand relationships across documents instead of relying on file names or folder structures. Additional AI capabilities include recruiter-ready profile summaries, skill extraction, project analysis, career insights, and evidence-backed knowledge representation. The project follows a modular architecture, making it easy to extend with integrations such as GitHub, LinkedIn, cloud storage, and learning platforms. For the best evaluation experience, please use a valid Anthropic API key and upload a diverse set of sample documents to demonstrate the complete AI-powered digital identity workflow from document ingestion to intelligent retrieval.

5 media files · digital-identity-system-tau.vercel.appView

EXPIRY ALERT ' 26

Setup Instructions: Requires Node.js 18+, PostgreSQL 14+, and an OpenAI API key Clone repo → cd server && npm install && cp .env.example .env → add your DATABASE_URL and OPENAI_API_KEY in .env npx prisma generate && npx prisma migrate dev --name init && npm run seed npm run dev (backend on port 5000) New terminal: cd client && npm install && cp .env.example .env && npm run dev Test credentials / demo data: The seed script auto-loads 12 realistic records (expired, expiring soon, active) across 4 stakeholders — the dashboard is fully populated on first load. Design decisions: Status (Active/Expiring Soon/Expired) is purely deterministic date math — never AI-derived, so compliance logic is always trustworthy Every AI call has a rule-based fallback so the app never breaks if the API is slow The chatbot is grounded in a live snapshot of the document DB — it cannot hallucinate records that don't exist

expiry-alert.onrender.comView

Smart Healthcare Appointment & Medical Records Platform

Millions of patients struggle to find suitable doctors, book appointments quickly, and maintain medical records in one place. Clinics often rely on phone calls, paper registers, or messaging apps, causing scheduling conflicts and poor record management. Doctors lack a centralized view of appointments and patient history. QueueCure was built to provide doctor discovery, secure appointment booking, health summaries, consultation notes, prescriptions, and reliable backend-level double-booking prevention. Process I started by analyzing the patient and doctor journeys to identify common pain points such as fragmented records, appointment conflicts, and inefficient scheduling. The platform was designed with two separate workflows: a patient portal for doctor discovery and booking, and a doctor dashboard for appointment management and consultation updates. MongoDB transactions were implemented to prevent double-booking under concurrent requests. Mobile responsiveness was prioritized to ensure bookings could be completed in under two minutes. Iterative testing was conducted for authentication, slot management, prescriptions, health summaries, and concurrent booking scenarios. Results The platform enables patients to search doctors, view availability, and complete bookings in under two minutes. Doctors can manage schedules, review patient health summaries, record diagnoses, and generate prescriptions from a single dashboard. Backend transaction handling ensures only one booking succeeds when multiple users attempt to reserve the same slot. The system successfully integrates appointment management, health records, and scheduling into a unified healthcare platform. Reflection Given more time, I would integrate telemedicine features, automated reminders through SMS and email, AI-assisted doctor recommendations, and electronic health record interoperability. I would also add advanced analytics for doctors, support multi-clinic operations, and conduct larger-scale usability testing with real patients and healthcare professionals to further optimize booking efficiency and user experience.

11 media files · ambula-healthcare-platform-qcen.vercel.appView

CampusCash AI – Intelligent Student Expense Tracker & Financial Assistant

College students make frequent small payments through UPI, canteens, travel, subscriptions, and mobile recharges, making it difficult to track where their money is being spent. Most expense-tracking solutions require mobile apps, internet access, or complex setup, creating friction for students who need a quick and lightweight solution. The goal was to build an offline expense tracker that not only records expenses but also provides intelligent financial insights such as spending forecasts, category analysis, affordability checks, and budget monitoring using only Python standard libraries. Process I started by implementing the core requirements: adding expenses, viewing expenses, calculating total spending, and identifying the highest spending category. After validating the basic workflow, I enhanced the system with budget management and spending alerts. To make the solution more useful, I introduced automatic category detection based on expense descriptions, month-end spending forecasts, affordability analysis, and spending insights. I initially used fixed categories but later redesigned the system to dynamically create new categories when users entered custom ones. The final solution focuses on simplicity, offline accessibility, and actionable financial recommendations rather than only storing transaction data. Results The final solution satisfies 100% of the required functionality while extending the project with several intelligent features. Users can track expenses, monitor budgets, forecast month-end spending, analyze spending patterns, and receive financial recommendations entirely offline. The system uses zero external dependencies and loads sample data immediately for demonstration. The addition of auto-categorization and financial insights transformed the project from a basic expense logger into a lightweight personal finance assistant suitable for students. Reflection With additional development time, I would integrate OCR-based receipt scanning to automatically extract expense information from bills and receipts. I would also replace the current keyword-based category detection with a machine learning model trained on real transaction descriptions to improve classification accuracy. Finally, I would add persistent storage using CSV files and build synchronization between the CLI application and the accompanying web dashboard for a more complete user experience.

9 media filesView

Live Digital Queue Manager for Clinics

Neighbourhood clinics in India still depend on paper token slips, verbal announcements, and manual queue tracking. Patients often wait 2-3 hours without knowing their position in the queue, leading to frustration and repeated inquiries at the reception. Receptionists must handle registrations, assign tokens, answer patient questions, and manage queue progression simultaneously. The goal was to build a digital queue management system that allows fast patient registration, provides live queue visibility on mobile devices, and generates realistic wait-time estimates using actual consultation data Process I designed the system around two user interfaces: a receptionist dashboard and a patient waiting-room view. To keep both screens synchronized, I initially explored periodic polling but found it introduced delays and unnecessary network requests. I switched to Socket.IO with WebSockets, enabling instant updates whenever a patient was added or a token was called. For wait-time estimation, I first used a fixed consultation duration, but the predictions became inaccurate when consultation lengths varied. To improve accuracy, I implemented an Exponential Moving Average (EMA) model that continuously learns from completed consultations. I then tested live synchronization, queue progression, priority handling, and wait-time calculations to ensure reliability and responsiveness. Results The system allows receptionists to register patients and assign tokens in under 10 seconds. Queue updates are synchronized instantly across all connected screens without page refreshes. Patients can view their token status, tokens ahead, and estimated wait times directly from their phones. Unlike traditional systems, wait estimates are generated using real consultation data through an adaptive EMA algorithm, making predictions more accurate as clinic activity changes throughout the day. Reflection With more time, I would add WhatsApp and SMS notifications so patients can leave the waiting area and receive alerts before their turn. I would also introduce cloud-based storage and analytics dashboards to help clinic owners track patient flow, peak hours, and doctor performance. Conducting usability tests with real receptionists and patients would provide valuable feedback for improving accessibility, workflow efficiency, and overall user experience.

10 media files · queue-cure.vercel.appView

This is suchitha’s work on Wooble.

Build a profile that shows what you can do — and share it anywhere.

Build yours