MemGallery is an AI-powered memory assistant Android app that democratizes brand-exclusive features like OPPO's Mind Space and Nothing's Essential Space, bringing automatic screenshot detection, multimodal capture, and semantic search to any Android 7.0+ device. Built with Jetpack Compose, Kotlin Coroutines, Hilt DI, Room, and Google's Gemini 2.5 Flash API, it automatically captures and processes screenshots, photos, audio, and text—extracting actionable insights (events, todos, reminders) and making everything searchable through natural language. The project showcases modern Android architecture (MVVM + Repository pattern),

Technology: Kotlin

Tags: Kotlin, Android, Jetpack Compose, Material Design 3, MVVM, Clean Architecture, Hilt, Dagger, Room Database, Coroutines, Flow, WorkManager, Gemini API, Generative AI, Google AI SDK, Background Processing, Mobile Development, Open Source

View Live Project

View Source Code

AIDA (AI-Driven Assistant) is a full-stack personal command center built with Next.js 15, MongoDB, and the Google Gemini API. It unifies task management, goal tracking, notes, and calendar events into a single dashboard. The core feature is an intelligent assistant that processes natural language commands (text and audio/multimodal) to execute CRUD operations, leveraging persistent user context and custom data models for dynamic goal calculation.

Technology: TypeScript

Tags: Next.js 15, React 19, TypeScript, MongoDB, Mongoose, NextAuth.js, Google Gemini API, AI/NLP, Tailwind CSS, shadcn/ui, Custom Theming, Multimodal Input, CRUD API

View Live Project

View Source Code

StudyInc is a comprehensive Electron desktop application that transforms collaborative learning. It integrates Pomodoro timers, real-time video calls, and task management within structured study sprints, enabling users to boost focus and productivity. Leveraging Next.js, Node.js, MongoDB, and Socket.IO, StudyInc provides a seamless, interactive environment for group study, complete with detailed analytics to track progress and optimize learning strategies.

Technology: TypeScript

Tags: ✦ JavaScript, TypeScript, Node.js, Next.js, React, Express.js, Electron, MongoDB, Socket.IO, WebRTC, Tailwind CSS, Collaborative Study, Pomodoro Timer, Video Calls, Real-time Chat, Task Management, Analytics Dashboard, User Authentication, Sprint Management, Desktop Application, IPC, Notifications, REST API, JWT, Mongoose, Simple Peer, PeerJS, React Query, Radix UI, shadcn/ui

View Source Code

An AI-powered web application built with React and TypeScript that converts static PDF files into interactive, editable study notebooks using the Google Gemini API. It features real-time Markdown streaming, intelligent content summarization, on-demand content expansion, searching, and custom quiz generation based on saved material, all managed client-side using local storage.

Technology: TypeScript

Tags: React, TypeScript, Google Gemini API, AI, PDF Processing, Markdown, Vite, Tailwind CSS, Client-Side Storage, Streaming, Quiz Generation, Highlight.js

View Live Project

View Source Code

A Q-learning implementation to solve package collection tasks in the classic Four-Rooms grid world environment. The agent learns optimal policies across three complexity levels: single-package collection, multi-package unordered collection, and strict sequential multi-package collection (RGB order), supporting both deterministic and stochastic environment dynamics.

Technology: Python

Tags: Python, Reinforcement Learning, Q-Learning, Grid World, NumPy, Matplotlib, Machine Learning

View Live Project

View Source Code

A high-performance B2B lead generation engine built on FastAPI and Google Gemini (GenAI SDK). It uses a two-step streaming AI pipeline (research and formatting) with tools (Google Search, URL Context) to create deeply analyzed, structured sales profiles and hyper-personalized outreach messages in real-time, streamed via Server-Sent Events (SSE) to a dynamic frontend UI.

Technology: HTML

Tags: FastAPI, Python, Google Gemini, GenAI SDK, Prompt Engineering, Server-Sent Events (SSE), Streaming, AI, Lead Generation, Pydantic, Jinja2, CSV, REST API, Tailwind CSS, HTML

View Live Project

View Source Code

A sophisticated, self-updating portfolio platform built on Next.js and TypeScript, featuring a secure administration dashboard that integrates the Google Generative AI (Gemini) API to automatically fetch code from GitHub repositories, analyze the codebase, and generate detailed, professional Markdown project descriptions for MongoDB storage.

Technology: TypeScript

Tags: Next.js, React, TypeScript, Tailwind CSS, NextAuth, MongoDB, Mongoose, Gemini API, AI Integration, Full-Stack, typescript

View Live Project

View Source Code

Nexa is a capstone project that addresses the critical issue of limited internet access in rural and township communities by providing a decentralized, offline-first mobile communication platform. Leveraging Delay-Tolerant Networking (DTN) principles, Nexa enables secure, end-to-end encrypted messaging, group chats, and multimedia sharing through device-to-device "message hopping" via Bluetooth and Wi-Fi Direct, ensuring communication resilience even without traditional internet infrastructure. This innovative approach empowers underserved populations with reliable connectivity, fostering digital inclusion and community engagement

Technology: Kotlin

Tags: ✦ DTN, Android, Kotlin, Jetpack Compose, P2P, Offline-First, Cryptography, Google Tink, Nearby Connections API, Mobile Development

View Source Code

Calculon is a real-time, conversational math chatbot powered by Google Gemini and a decoupled Modular Calculation Protocol (MCP) server. The system utilizes FastAPI and WebSockets to enable secure, streamed tool use, forcing the AI to delegate all arithmetic operations (add, subtract, multiply, divide) to a precise, dedicated backend service. The frontend displays the AI's thoughts, tool calls, and results in real time, showcasing advanced RAG/Tool-Use orchestration.

Technology: Python

Tags: Python, FastAPI, WebSockets, Google Gemini, Generative AI, Tool Use, Modular Calculation Protocol (MCP), Vanilla JavaScript, HTML, CSS, Real-time Streaming, Asynchronous Programming, Uvicorn, python

View Live Project

View Source Code

clia is an advanced, conversational AI assistant for software engineering, implemented as a command-line tool (CLI) and a web UI. Powered by the Google Gemini model, it uses a local Model Context Protocol (MCP) tool server to safely and transparently interact with the user's filesystem and command line. It features streaming AI thoughts, mandatory tool call permissions, and precise file manipulation capabilities, enabling developers to refactor, debug, and implement features through natural language commands.

Technology: Python

Tags: Python, Google Gemini, GenAI, LLM, Tool Use, MCP, rich, prompt-toolkit, FastAPI, CLI, Software Engineering, Asynchronous Programming, python

View Live Project

View Source Code

Verilookie is a full-stack web application that provides unified deepfake and AI-generated media detection, utilizing a Python FastAPI backend and a React frontend. It uniquely integrates a gamified learning system powered by the Google Gemini API to offer personalized, adaptive quizzes, educating users while protecting them from misinformation.

Technology: Python

Tags: Python, FastAPI, React, JavaScript, Deep Learning, Vision Transformer (ViT), Hugging Face, NVIDIA AI, Google Gemini, Docker, Gamification, Cybersecurity, python

View Live Project

View Source Code