# Android Projects at AI Tinkerers

> Canonical HTML: https://aitinkerers.org/technologies/android
> Markdown URL: https://aitinkerers.org/technologies/android.md
> Technology record last updated: 2026-02-22T17:59:34Z
> Generated: 2026-08-25T13:13:18Z

The Google-developed, Linux-based, open-source mobile operating system powering billions of devices globally.

Android is the world's dominant mobile operating system, developed by Google and built on the Linux kernel. First released in 2008, it quickly established its open-source platform (AOSP), enabling massive hardware diversity across OEMs (Samsung, Xiaomi, etc.). The platform currently commands a significant global mobile OS market share, serving over three billion monthly active users. Its reach extends beyond smartphones and tablets: dedicated versions power smart TVs (Android TV), cars (Android Auto/Automotive), and smartwatches (Wear OS).

- Official technology site: https://android.com
- Public AI Tinkerers demos and talks: 11
- Result page: 1 of 1

## Recent Public Talks and Demos

### [UnaMentis: A mobile, AI, voice first learning platform](https://portland.aitinkerers.org/talks/rsvp_L4vjfVilvoQ)

Covering our entire project is very much out of scope, but I will be giving a quick background on who we are and what we are building. The primary topic though is our very aggressive use of on de vice voice models to provide both the lowest latency and the highest quality in a detached client we can. This includes porting a TTS model that was a perfect fit for our use to IOS with rust/candle.

- Event context: AI Tinkerers Portland: Building Voice Agents & Conversational AI Stacks — 2026-03-06 — Portland
- Public talk page: https://portland.aitinkerers.org/talks/rsvp_L4vjfVilvoQ

### [PicoClaw on MaixCam and OpenClaw on Android: Making a Claw Machine](https://manizales.aitinkerers.org/talks/rsvp_ltSlhnfnf4Y)

This is a technical teardown of how I got PicoClaw running on a MaixCam (K210-based) board and OpenClaw on Android client, turning a claw into a live phone hardware controller No slides. Just code, firmware, serial logs, and hardware. The demo will cover: 1. Porting PicoClaw to MaixCam PicoClaw was designed with a specific embedded environment in mind. I’ll walk through: Adapting the firmware to run on MaixCam Fixing build system assumptions Debugging boot and runtime issues We’ll look at: What broke immediately What silently failed What needed to be stubbed or reworked Timing quirks and hardware assumptions in the original firmware This section is mostly about embedded friction and reality. 2. Running OpenClaw on Android On the Android side, we’ll walk through: USB permission handling Threading model Command dispatch UI → Command translation We’ll examine: How commands are built How the app handles connection drops What assumptions the Android client makes about the firmware Where the abstraction leaks This is not a UI demo — it’s a transport-layer autopsy.

- Event context: 🚀 ¡13vo Encuentro de AI Tinkerers Manizales! 🤖 — 2026-02-26 — Manizales
- Public talk page: https://manizales.aitinkerers.org/talks/rsvp_ltSlhnfnf4Y

### [Using AI to channel Olympic excitement into skill development of homegrown athletes](https://toronto.aitinkerers.org/talks/rsvp_pFLrl0ONOj0)

I coach at a speedskating club, and one of the most difficult aspects of training young athletes is focusing them on technical aspects, tracking progress, and communicating tips and form correction effectively. This is a mobile app designed specifically with speedskaters in mind, and it gamifies the process of holding a "basic position" i.e. the crouched-over skating position. The user begins a game and video is captured. A wireframe is overlaid on the athlete, and the knee and hip angles are calculated. When the athlete goes into basic position, the user enters a session and begins to accrue a score. Score accrual multipliers are applied the closer to "perfect form" an athlete has, which has visual indicators displayed in realtime. Feedback from an AI coach is also given in realtime, which keeps the user engaged and focused. The user finishes a session after they stand up, and a report generated by AI is shown to help them understand what could be improved, as well as their final score. AI also uses the user's home country to query a public API on speedskater results and world records (ex. a Canadian user would be shown world records and stats of famous skaters such as Laurent Dubreuil). It helps users connect more with how strong the Olympic-level athletes are, along with recording progress and showing improvement over time.

- Event context: AI Tinkerers Toronto - February 2026 @ Cohere! — 2026-02-26 — Toronto
- Public talk page: https://toronto.aitinkerers.org/talks/rsvp_pFLrl0ONOj0

### [AI agents as interface layer for mobile apps](https://tiruchirappalli.aitinkerers.org/talks/rsvp_qrJiLeqgTOo)

Mobile interfaces are reaching their limit. More features mean more screens, more taps, and more friction. This talk explores a different direction. AI agents not as assistants bolted onto apps, but as the interface itself. Through a live demo of Kuralit, I show how user intent can bypass traditional UI and directly trigger real actions inside an app. No buttons. No navigation trees. Just intent to execution. This is not about voice for convenience. It is about interface evolution. From visual control to intent-driven software. The goal is to question a simple assumption. If software can understand what a user wants, why does it still wait for clicks?

- Event context: AI Tinkerers Trichy: January Meetup & Live Demos — 2026-01-31 — Tiruchirappalli
- Public talk page: https://tiruchirappalli.aitinkerers.org/talks/rsvp_qrJiLeqgTOo

### [What is an effective AI model for dolphin communication research?](https://montreal.aitinkerers.org/talks/rsvp_ldL1kjb_pss)

I am presenting a working Android app that can be used to embed one or more TFLite models for improved effectiveness. The models need to be developed. Current GenAI models are not a good fit because the data to train them does not exist and their use of image recognition on spectrograms of sounds is not optimal. One tech being considered is the one presented in kyutai.org .

- Event context: AI Tinkerers Montreal: Demo Night — November 20, 2025 — 2025-11-20 — Montreal
- Public talk page: https://montreal.aitinkerers.org/talks/rsvp_ldL1kjb_pss

### [Quantization for Edge AI](https://nairobi.aitinkerers.org/talks/rsvp_ax9_20x3YMo)

So, I’ll be presenting my journey building my first Android APK, CHWs Augment. The focus will be on augmentation, because with the rise of AI, people often think of it as a replacement tool—but my project explores AI as an augmentor, designed to enhance human work rather than replace it. Technically, the project dives into quantization, Edge AI, and responsible AI, with an emphasis on equity: How can someone in a remote area benefit from AI just like someone in the city? How can we build free AI solutions that run efficiently on low-budget phones and scale augmentation to improve efficiency for CHVs? I’ll be doing a live demo of the APK, showing its workflow and the tangible ways it helps CHVs in remote areas. The focus is on impact, not revenue: building scalable AI that improves work and accessibility. Along the way, I’ll share the internals—the code, model design, and deployment choices—so the audience sees not just what it does, but how it was built. This project is imperfect and experimental, which is intentional: sharing raw builds accelerates learning, sparks collaboration, and pushes the boundaries of AI deployment in resource-constrained settings.

- Event context: AI Tinkerers Nairobi - Edition #4 (Nov 6, 2025) — 2025-11-06 — Nairobi
- Public talk page: https://nairobi.aitinkerers.org/talks/rsvp_ax9_20x3YMo

### [A platform to embed your model to chat with dolphins at sea](https://montreal.aitinkerers.org/talks/rsvp_Sc0UeVyAbLg)

A working Android application in which a model developer could embed their own TFLite model. Possible functions: 1) filter noise, 2) recognize complex vocalizations, 3) propose acoustic replies.

- Event context: AI Tinkerers Montreal – Spooky October 2025 Meetup — 2025-10-21 — Montreal
- Public talk page: https://montreal.aitinkerers.org/talks/rsvp_Sc0UeVyAbLg

### [Offline AI Tutor on Android: Gemma 3n + MediaPipe GenAI with Local RAG+CAG implementation](https://seattle.aitinkerers.org/talks/rsvp_8B20pGxfZiw)

I’ll live-demo an offline AI tutor running entirely on an Android phone using Gemma 3n via MediaPipe. The app ingests local text files, chunks and semantically indexes them in SQLite, and performs retrieval-augmented generation with on-device embeddings. Please suggest what from code would be interesting to present..

- Event context: AI Tinkerers Seattle September Meetup — September 30, 2025 — 2025-10-01 — Seattle
- Public talk page: https://seattle.aitinkerers.org/talks/rsvp_8B20pGxfZiw

### [AI Game Master](https://seattle.aitinkerers.org/talks/rsvp_qGMmUEDayLk)

Our AI driven interactive adventure game - AI Game Master - is available on iOS and Android, and players really love it! We're completely bootstrapped and crossed the 100,000 players by now!

- Event context: AI Tinkerers Seattle - January 2025 Meetup — 2025-01-23 — Seattle
- Public talk page: https://seattle.aitinkerers.org/talks/rsvp_qGMmUEDayLk

### [Dynamic Structured Record Augmentation of LLMs](https://nyc.aitinkerers.org/talks/rsvp_o5c_H5O-V5U)

An approach that enables LLMs to dynamically create, manipulate, and retrieve structured records in a NoSQL database. This approach serves as an alternative to vector-based RAG, creating a data store that's both human and model-readable. Demoed in a mobile app for tracking daily records (e.g., cat medication, exercise, expenses) to illustrate practical applications. The core mechanism I’m focusing on is prompting for structured Tool/Function completions to do CRUD operations on structured records - to capture, remember, retrieve, and query against information the user wants to reference later. The schema / structure of the records themselves is also frictionlessly created / edited by the model. Both the table structures and the records themselves are stored in a NoSQL database. The “magic” is in actively managing the system prompt on the chat chain to provide full context and dynamically generate the functions available as tools for each completion. This “dynamic but structured” approach uses JSON Schema to define, communicate, and validate functions. A technical challenge was dynamically generating and validating table / context-specific functions. I did this by embracing JSON Schema in the table definition as well as the functions presented to the LLM. I’ll focus my demo on the prompt, the Firestore database, and the function building. This is a Flutter app that runs on iOS and Android. I’m very excited to share this approach with the other Tinkerers - shown as a fully working implementation that can be tried out, as well as access to all the code.

- Event context: AI Tinkerers July Meetup — 2024-07-24 — New York City
- Public talk page: https://nyc.aitinkerers.org/talks/rsvp_o5c_H5O-V5U

### [Building Multi-Agent Generative-AI Applications with AutoGen](https://palo-alto.aitinkerers.org/talks/rsvp_NlYdJI4dowg)

Generative AI models have made significant progress on tasks such as summarizing passages, extracting entities and generating code etc. However, they currently struggle to address more complex tasks that require multi-step planning, reasoning, and action - for example, building a complete Android app for displaying stock prices. To create helpful AI assistants that can seamlessly handle these complex tasks, it is crucial to develop agents that can act, collaborate with other entities (including humans), and serve as interfaces to the digital world. But how do we define such multi-agent workflows, empower developers to build them, and address the open challenges that arise? AutoGen is a pioneering attempt to answer these questions and offers a generic framework for building multi-agent AI applications. In this talk I will briefly introduce the AutoGen framework and show a demo of AutoGen Studio - a low code interface that enables rapid prototyping of multi-agent applications. AutoGen is an open-source project (MIT License) on GitHub - https://github.com/microsoft/autogen. With over 24k stars, more than 250 contributors, and a growing ecosystem of integrations, AutoGen is a vibrant community of enthusiasts prototyping applications across various industry use cases.

- Event context: AI Tinkerers Palo Alto - Inaugural Meetup — 2024-05-01 — Palo Alto
- Public talk page: https://palo-alto.aitinkerers.org/talks/rsvp_NlYdJI4dowg

## Related Technologies

- [iOS](https://aitinkerers.org/technologies/ios) ([Markdown](https://aitinkerers.org/technologies/ios.md)) — 7 public demos
- [Java](https://aitinkerers.org/technologies/java) ([Markdown](https://aitinkerers.org/technologies/java.md)) — 8 public demos
- [Kotlin](https://aitinkerers.org/technologies/kotlin) ([Markdown](https://aitinkerers.org/technologies/kotlin.md)) — 6 public demos
- [AI Game Master](https://aitinkerers.org/technologies/ai-game-master) ([Markdown](https://aitinkerers.org/technologies/ai-game-master.md)) — 1 public demo
- [ASR](https://aitinkerers.org/technologies/asr) ([Markdown](https://aitinkerers.org/technologies/asr.md)) — 2 public demos
- [Autodesk Maya](https://aitinkerers.org/technologies/autodesk-maya) ([Markdown](https://aitinkerers.org/technologies/autodesk-maya.md)) — 1 public demo
- [Autogen](https://aitinkerers.org/technologies/autogen) ([Markdown](https://aitinkerers.org/technologies/autogen.md)) — 6 public demos
- [Autogen Studio](https://aitinkerers.org/technologies/autogen-studio) ([Markdown](https://aitinkerers.org/technologies/autogen-studio.md)) — 2 public demos
- [BERT](https://aitinkerers.org/technologies/bert) ([Markdown](https://aitinkerers.org/technologies/bert.md)) — 179 public demos
- [Blender](https://aitinkerers.org/technologies/blender) ([Markdown](https://aitinkerers.org/technologies/blender.md)) — 4 public demos
- [BLOOM](https://aitinkerers.org/technologies/bloom) ([Markdown](https://aitinkerers.org/technologies/bloom.md)) — 115 public demos
- [DreamFusion](https://aitinkerers.org/technologies/dreamfusion) ([Markdown](https://aitinkerers.org/technologies/dreamfusion.md)) — 1 public demo
- [Edge processing](https://aitinkerers.org/technologies/edge-processing) ([Markdown](https://aitinkerers.org/technologies/edge-processing.md)) — 1 public demo
- [Embedded C/C++](https://aitinkerers.org/technologies/embedded-c-c) ([Markdown](https://aitinkerers.org/technologies/embedded-c-c.md)) — 2 public demos
- [Firestore](https://aitinkerers.org/technologies/firestore) ([Markdown](https://aitinkerers.org/technologies/firestore.md)) — 4 public demos
- [Flutter](https://aitinkerers.org/technologies/flutter) ([Markdown](https://aitinkerers.org/technologies/flutter.md)) — 22 public demos
- [Fusion 360](https://aitinkerers.org/technologies/fusion-360) ([Markdown](https://aitinkerers.org/technologies/fusion-360.md)) — 1 public demo
- [GeckoEmbeddingModel](https://aitinkerers.org/technologies/geckoembeddingmodel) ([Markdown](https://aitinkerers.org/technologies/geckoembeddingmodel.md)) — 1 public demo