Neshandra G

B.Tech IT Student | Passionate About AI | Full-Stack Developer

Chennai Institute of TechnologyChennai, TamilNadufull_time, internship

Open to roles

GITProject ManagementProblem SolvingLeadership

Helix — AI Digital Identity System

HELIX is an AI-powered Digital Identity & Knowledge Graph that transforms scattered academic and professional documents into a structured, searchable, and evidence-backed representation of a student's growth. Instead of simply storing files, HELIX understands uploaded content, automatically categorizes information, discovers relationships across experiences, visualizes growth through a digital journey timeline, and enables natural-language retrieval using Retrieval-Augmented Generation (RAG). PROJECT RESOURCES Live Application https://h-e-l-i-x-peach.vercel.app Demo Video https://youtu.be/5ONEuIMZQLc Presentation https://drive.google.com/file/d/1lZKq4jQd_KRlNsp7IG7yQ0Dzpa9iz_Vp/view?usp=sharing GitHub Repository https://github.com/neshandrag/h.e.l.i.x EXPLORING - HELIX • Register a new account (no test credentials are required). • Upload certificates, resumes, project reports, internship letters, or images. • Optionally connect a GitHub repository to import project information automatically. • Explore the Documents Dashboard to review uploaded files, AI classifications, and evidence scores. • Visualize relationships between skills, projects, certifications, internships, and achievements using the Knowledge Graph. • View your milestones in the Digital Journey Timeline, automatically generated from uploaded evidence. • Use Ask AI to query your digital identity in natural language and receive evidence-backed responses powered by Retrieval-Augmented Generation (RAG). • Visit the Public Profile to view a shareable, read-only representation of your digital identity. KEY DESIGN DECISIONS • AI is responsible for information extraction and categorization, while verifiability, relationship depth, and path coherence are computed using deterministic algorithms for transparency and consistency. • Classification, Verifiability, Relationship Depth, and Coherence are evaluated independently instead of being combined into a single opaque score. • Semantic search powered by vector embeddings and Retrieval-Augmented Generation (RAG) ensures responses are grounded in uploaded evidence rather than generated assumptions. • Original documents remain preserved in their native format and are always accessible. • The modular architecture allows document uploads to function independently, while GitHub and Telegram integrations remain optional extensions. DEPLOYMENT INFORMATION Frontend: Vercel Backend: Render Database: Supabase (PostgreSQL + pgvector) Note: The backend is deployed on Render's free tier. After periods of inactivity, the first API request may take approximately 30–60 seconds while the service resumes. Subsequent requests respond normally.

5 media files · h-e-l-i-x-peach.vercel.appView

Quro : Real-Time Digital Queue Management System

Most neighborhood clinics still rely on paper tokens and verbal announcements. Patients often wait 2–3 hours without knowing their queue position or expected wait time, while receptionists repeatedly answer the same questions and manually manage patient flow. This creates frustration, inefficiency, and poor visibility for both patients and staff. The goal was to build a real-time digital queue system that provides live tracking, accurate wait estimates, and instant updates across reception, waiting-room displays, and patient phones. Process I analyzed queue management workflows in small outpatient clinics and identified three key challenges: poor patient visibility, repetitive receptionist workload, and disconnected communication across clinic systems. Based on these findings, I designed a unified three-interface solution. After evaluating a polling-based approach, I transitioned to Socket.io for real-time synchronization. Reliability was enhanced through REST fallback polling, MongoDB recovery, emergency prioritization, and no-show handling. Iterative improvements focused on faster registration, accurate wait-time estimation, and improved accessibility through QR access, multilingual support, voice guidance, and an AI-powered assistant. Results Quro achieved its core goals with under 10-second patient registration and sub-second queue synchronization across reception, waiting-room displays, and patient phones. Wait times are calculated from real consultation data, improving accuracy throughout the day. The platform supports 6 Indian languages, QR-based access, AI-assisted queries, and automatic session recovery. Testing confirmed reliable handling of emergencies, no-shows, and concurrent registrations. Planned enhancements include SMS notifications, appointment scheduling, analytics, and real-world clinic deployment. Reflection Building Quro showed me that creating a successful product involves more than solving the technical challenge. I would place greater emphasis on gathering feedback from clinic staff and patients earlier to refine workflows and improve accessibility. I would also add SMS and WhatsApp notifications, enhance wait-time prediction, and expand the platform with appointment scheduling, doctor-facing tools, and analytics to create a more complete clinic management solution.

11 media files · quroclinic.vercel.appView