MANSI — Where Emotion Meets Understanding

A privacy-first digital mental wellness platform connecting students and counselors through anonymous support and real-time wellbeing intelligence.

Comment

💡 Inspiration

In competitive academic environments, over 70% of students facing severe burnout and emotional distress never seek help. The primary barrier is not a lack of counselors, but the fear of social stigma, privacy leaks, and clunky institutional barriers that make students feel digitally and emotionally isolated.

As final-year engineering students experiencing this high-pressure environment firsthand at ABV-IIITM Gwalior, our team led by Mansi Singh built MANSI (Mental Awareness, Nurturing & Support Interface)—a holistic, privacy-preserving digital sanctuary that bridges the gap between students, counselors, teachers, and university leadership.

✨ What it does

MANSI connects four campus stakeholders into one cohesive, privacy-preserving network:

  1. Student Portal (Zero-Stigma Anonymity): 1-on-1 live encrypted chat with verified psychologists using ephemeral aliases, 30-second clinical wellness check-ins, and real-time AI triage support.
  2. Psychologist Portal: Prioritized triage inbox of incoming anonymous sessions with category tagging (Exam Anxiety, Social Isolation) without unmasking student identity.
  3. Teacher Portal: Classroom attendance and academic stress correlation to detect early drop-off signals.
  4. Institutional Administration: Zero-knowledge aggregate stress radar powered by dynamic Recharts visualizations to monitor campus-wide mental health trends without tracking individuals.

🛠️ How we built it

  • Frontend & App Framework: Next.js 14 (App Router), React 18, TypeScript, Tailwind CSS, Lucide Icons.
  • Data Visualization & Dashboards: Recharts (Dynamic BarCharts, AreaTrends, and Radial Stress Radars).
  • Backend & Serverless Layer: Next.js Serverless Route Handlers (/api/auth, /api/chat, /api/institute, /api/student, /api/ai-chat).
  • Database & Identity Isolation: MongoDB with Mongoose ODM modeling decoupled schemas (User, ChatSession, Message, RiskProfile, AcademicRecord).
  • Authentication & Security: Stateless JSON Web Tokens (JWT), bcryptjs encryption, secure HTTP-only cookie sessions (cookies-next).
  • Data Processing: PapaParse & Formidable for fast CSV data ingestion and anonymization.
  • Deployment: Vercel & MongoDB Atlas.

🔄 What we changed (Mandatory Mentor Feedback Iteration)

"During our Saturday mentor check-in, our mentor observed that our initial triage questionnaire forced students through 6 multi-step screens before allowing them to initiate a confidential chat with a counselor, causing high abandonment during acute anxiety moments. Taking this advice to heart, we refactored the entire flow to implement a 1-tap 'Instant Safe Connect' launcher that connects students to an anonymous room immediately, shifting detailed assessments into an optional post-chat reflection. This decreased time-to-first-connection from over 100 seconds to under 8 seconds."

🧗 Challenges we ran into

  • Zero-Knowledge Decoupling: Designing a database schema where administrators can verify student enrollment without possessing any capability to link real identities to private counseling transcripts.
  • Role-Based Access Control (RBAC): Managing seamless stateless JWT routing across 4 distinct user roles in Next.js 14 App Router.
  • Dynamic Visual Analytics: Aggregating clinical risk factors into responsive, mobile-friendly Recharts graphs.

🏆 Accomplishments that we're proud of

  • Delivering a fully working 4-role platform with live MongoDB persistence in under 48 hours.
  • Building an empathetic, soothing UX design that feels comforting rather than clinical.
  • Formulating an institutional analytics model that provides actionable wellness intelligence while fiercely defending student privacy.

📚 What we learned

  • Designing for vulnerable emotional contexts requires extreme friction reduction and trauma-informed UI palettes.
  • Architectural patterns for decoupled, privacy-preserving microservices in Next.js 14.
  • The importance of listening to mentor feedback to eliminate feature bloat.

🔮 What's next for MANSI

  • WebRTC Encrypted Audio/Voice: Adding voice distortion for anonymous voice counseling.
  • Predictive Burnout Modeling: Edge ML models alerting leadership 2 weeks before major exam burnout spikes.
  • LMS Integrations: Open Canvas/Moodle plugins for privacy-safe SSO.

Built With

Share this project:

Updates