Android apps, AI agents, on-device LLMs: built, broken, rebuilt, and shipped. Every project comes with receipts: a full writeup, a demo video, and the GitHub repo.
A small live demo comparing Google's DiffusionGemma against regular Gemma 4 head to head: paste a paragraph with a planted mistake, watch both models fix it live, and see exactly how many words each one actually touched.
Give it any GitHub repository URL. Three AI agents built on three different frameworks — Google ADK, LangGraph, and CrewAI — communicate with each other using the Agent-to-Agent (A2A) protocol to produce a complete developer onboarding kit: a plain-English guide, a data flow diagram, and the five files every new developer should read first.
Telegram bot powered by Google ADK 2.1 and Gemini 2.5 Flash that extracts your Reading DNA from books you love, then recommends your perfect next read with a match score, book cover, and follow-up options. Built with FastAPI, deployed on Google Cloud Run.
Full-stack web app that uses Gemini Embedding 2 to convert photos, PDFs, text files, and audio into 3,072-dimensional vectors, then searches across all of them by meaning. Type a sentence and find a photo, a text note, and a PDF on the same topic. Built with FastAPI, Next.js, and ChromaDB.
Android app powered by 4 AI agents, 3 running simultaneously in parallel, that researches any product across Google Shopping, Reddit, and YouTube, then synthesizes everything into one ranked recommendation with clickable links. Built on Google ADK and deployed to Vertex AI Agent Engine.
Android app that photographs recipes in 140+ languages and uses Gemma 4 via LiteRT-LM to translate and structure them into ingredients, steps, and cooking time. Everything runs on the phone with no internet needed.
Android module that kicks in after a crash is detected by another app. Uses a local Gemma LLM to assess driver injury through guided voice conversation and sends emergency SMS automatically if the driver stops responding.
Android PoC that generates professional FNOL (First Notice of Loss) insurance reports by combining simulated telematics data with real-time location, weather, and traffic context, processed on-device via MediaPipe Gemma.
Evolution of AIDrivingHelper that enriches driving events with real-world geographic context, including nearby schools, hospitals, road names, and traffic infrastructure, for location-aware safety coaching on Android.
Android module that uses RAG entirely on-device to answer natural language questions about past trips and deliver personalized driving coaching. Built to plug into apps that already record trip data, with voice input and voice output.