Portfolio

Selected Work

Production-grade AI systems, tools, and experiments.

Platform

Roles

3-tier RBAC

Platform

Flutter

TowerAssist

Cross-platform building maintenance and operations platform.

A Flutter app (Android, iOS, Web) for building operations. Residents raise maintenance tickets, technicians manage and close them, admins oversee everything — with real-time Firestore, FCM push notifications, 3-tier RBAC via custom Firebase claims, audit logging, and clean feature-first architecture.

FlutterFirebaseRiverpod

Lesson: Real-time role-based platforms need clean separation from day one.

AI Tool

AI Model

LLaMA 3.3 70B

Privacy

Encrypted keys

AI Study Planner

Privacy-first AI study planner with voice, OCR, and mock tests.

A React Native study app powered by Groq LLaMA 3.3 70B. Scan a syllabus via OCR (Apple Vision / ML Kit + Groq Vision fallback), speak tasks with Groq Whisper, generate AI mock tests with answer evaluation, and get personalised study sessions — all with per-user encrypted API keys and a prompt injection guard.

React NativeGroq (LLaMA 3.3)Expo

Lesson: Privacy-first AI: user-owned keys, on-device protection.

Education

Approach

Narrative

Format

Interactive

Panchatantra Algorithm Series

Storytelling method for teaching algorithms.

A content + code project that uses narrative storytelling to teach complex algorithms. Each algorithm is a "story" with characters, conflict, and resolution — making concepts stick.

Ghost CMSAlgorithm DesignInteractive Storytelling

Lesson: Teaching through narrative.

AI Assistant

Knowledge

Indian Law RAG

Input

Voice + Text

LegalAI

AI legal assistant for Indian law, built with Flutter and Google Gemini.

A Flutter app that makes Indian legal knowledge accessible through conversation. Built-in RAG knowledge base covering tenant rights, consumer protection, traffic law, and FIR procedures. Uses Google Gemini with dynamic model selection, speech-to-text voice input, and markdown-rendered responses.

FlutterGoogle GeminiSpeech-to-Text

Lesson: Domain-specific RAG makes specialist knowledge accessible to everyone.