# Amir

Amir is a design/product engineer working at the intersection of AI and human-computer interaction. Since 2016 his work has been about building software that adapts to people instead of asking them to adapt to it — across gesture recognition systems, semantic search engines, AI developer infrastructure, adaptive memory layers, and now interfaces for AI agents.

## Projects
- [Serene](https://amir.cloud/projects/serene.md) — A sealed desktop appliance and local-first software system that serves as a private memory aid for therapists, emphasizing physical privacy through visible hardware rather than policy.
- [Adapt](https://amir.cloud/projects/adapt.md) — Open-source Node.js library implementing self-organizing memory systems with Brain orchestration and specialized Neurons that compress understanding and autonomously reconfigure based on detected gaps.
- [Agent Mind](https://amir.cloud/projects/agent-mind.md) — Agent Mind provides persistent shared memory for AI coding agents, enabling Claude Code, Cursor, and Codex sessions to start with accumulated context through a CLI-driven MCP server.
- [TOOK.wiki](https://amir.cloud/projects/took-wiki.md) — TOOK.wiki is an online reader generator that extracts entities from uploaded texts and contextualizes them through Wikipedia to create new publishable narratives.
- [Unbody](https://amir.cloud/projects/unbody.md) — Open-source AI infrastructure that uses enhancer agents to automatically augment data—crawling URLs, captioning images, transcribing video—before exposing retrieval, memory, and generative search through a single API.
- [Gray](https://amir.cloud/projects/gray.md) — An open-source AI-native blogging framework enabling semantic search, natural language queries, and conversational exploration of blog content.
- [HIRO](https://amir.cloud/projects/hiro.md) — HIRO is a marketplace where generative artworks expose tunable parameters through an SDK, auto-generating editor controls that let buyers customize and export unique renditions in real time.
- [Autonomous Libraries](https://amir.cloud/projects/autonomous-libraries.md) — R&D project reimagining libraries as distributed, community-run spaces where physical books connect to their digital context through semantic search, gesture interaction, and light AR.
- [IIA (Intelligence Inventory Assistant)](https://amir.cloud/projects/iia-intelligence-inventory-assistant.md) — Led product engineering for a camera-first mobile app that recognizes objects to manage collections, shipping v1.0 on iOS and Android in 2022.
- [Intention Recognition](https://amir.cloud/projects/intention-recognition.md) — A gesture recognition system combining skeleton tracking, facial analysis, and object detection to let users control software through natural movements and object handling.
- [Knowledge Recognition](https://amir.cloud/projects/knowledge-recognition.md) — Led development of a 2018 semantic search system combining YOLOv3, Mask R-CNN, Tesseract OCR, and custom Word2Vec embeddings to enable idea-based search across document collections.
- [OPENRNDR.org](https://amir.cloud/projects/openrndr-org.md) — Website for OPENRNDR, an open-source creative coding framework, featuring documentation and project showcase.
- [Pixelator](https://amir.cloud/projects/pixelator.md) — A DrawBot module that pixelates text, created as an early experiment in building custom generative typography tools.
- [Tomorrow Book](https://amir.cloud/projects/tomorrow-book.md) — An experimental site that generates new reading narratives for books from Wikipedia articles, exploring what digital screens offer beyond the printed page.
- [Text & Space](https://amir.cloud/projects/text-space.md) — An installation system that transforms text into navigable 3D space, plotting each word as a point and connecting them into an environment viewers can move through.
- [1 week of my life / publication](https://amir.cloud/projects/1-week-of-my-life-publication.md) — An early generative book transforming one week of daily activity logs into visual artworks, incorporating subjective time perception as a data dimension.
- [2001 a space odyssey / poster series](https://amir.cloud/projects/2001-a-space-odyssey-poster-series.md) — An early poster series using a camera filming its own screen output to create self-generating typographic designs for 2001: A Space Odyssey.
- [noui](https://amir.cloud/projects/noui.md) — Early machine learning experiments using StyleGAN2 and Pix2Pix GANs to explore whether user interfaces could be generated and personalized rather than hand-designed.
- [Project Indefensible](https://amir.cloud/projects/project-indefensible.md) — An educational website on the global arms business designed as a collaborative tool for reading and writing text together across different audiences and contexts.
- [ISIA Urbino Digital Platform](https://amir.cloud/projects/isia-urbino-digital-platform.md) — A publishing platform that transforms Google Drive, Docs, Sheets, and Calendar into a headless CMS serving content through REST and GraphQL APIs.
- [The 25th Hour](https://amir.cloud/projects/the-25th-hour.md) — A WebGL map visualization built at RNDR showing how autonomous cars and ride sharing could reshape traffic patterns in Ingolstadt, Germany.
- [Re-](https://amir.cloud/projects/re.md) — A graduation project creating a modular web browser that let users rebuild webpages from only the content modules they selected, producing lighter and quieter browsing experiences.

## Posts
- [Adaptation Is All We Need](https://unbody.io/blog/adapt) — Explains the motivation behind building Adapt, a memory system that synthesizes information and restructures itself as understanding evolves rather than just storing and retrieving data.
- [The End of the Container Era: Apps Are a Historical Glitch](https://unbody.io/blog/end-of-container-era-apps-glitch) — Essay arguing that apps are historical artifacts of distribution constraints, not natural thought patterns, and that future computing should abandon the container model for thread-based information flow.
- [Managing the UI state for generative data flow](https://unbody.io/blog/gen-data-ui-state) — Explores state management strategies for generative AI interfaces where outputs stream incrementally and fail unpredictably unlike traditional CRUD operations.
- [From Query to Context: Building Agentic RAG with Unbody's JSON Endpoint](https://unbody.io/blog/query-to-context) — A technical guide on building agentic RAG systems that parse user queries into structured parameters using Unbody's JSON endpoint for improved retrieval.
- [AI-native vs AI-forced](https://unbody.io/blog/ai-forced-vs-ai-native) — Examines how fragmented AI tooling forces developers into parallel ecosystems and contrasts this with a unified stack approach.
- [AI is Complex, Do Not Try This In-House](https://unbody.io/blog/ai-is-complex) — Practical guidance for founders on the underestimated challenges of adding AI to products and where non-ML engineers should realistically start.
- [What the F**k is an AI-native app?](https://unbody.io/blog/ai-native-app) — Explores what distinguishes truly AI-native applications from traditional apps with AI bolted on, arguing that users now want answers rather than information.
- [Unbody: all AI buzz in one endpoint, one line of code](https://unbody.io/blog/hello-world) — Unbody provides a headless AI API that creates a seamless modular pipeline connecting private data sources to frontend applications through a single endpoint.

## Talks
- [Unbody demo dec 2024 - the first AI-native development stack](https://www.youtube.com/watch?v=PSPAFVfIv-A) — A December 2024 demonstration showcasing projects built on Unbody, featuring AI feature additions to existing products and AI-native applications.
- [AI is complex — Don't try this in-house!](https://www.youtube.com/watch?v=swDatZwbZQs) — A talk guiding startup founders through the realities, challenges, and starting points for adding AI capabilities to their products.
- [What the F**k is an AI-native app and how to build one.](https://www.youtube.com/watch?v=-ypE9IidC4E) — A talk explaining what makes applications truly AI-native, demystifying RAG concepts and demonstrating practical approaches to building them.
- [AI Study Assistant Workshop trim](https://www.youtube.com/watch?v=P4Wx-nnChIA) — A two-hour workshop demonstrating how to build an AI study assistant using semantic search and rerankers for PDF and video content retrieval.
- [How I built an AI search over my messy desktop files in 4 minutes](https://www.youtube.com/watch?v=OikfiKbT48k) — A four-minute live demo building semantic AI search over messy local desktop files with a code walkthrough.
- [What is an AI-native app?](https://www.youtube.com/watch?v=0hx-iuLKdx0) — A React Dublin talk with live demo exploring AI-native app characteristics including chatbots, semantic search, and generative UI, arguing AI-first will become the default paradigm like mobile-first before it.
- [Unbody advanced AI for private data, from anywhere and in any formats, via one touchpoint.](https://www.youtube.com/watch?v=FP1NJuP7CQI) — Early talk introducing Unbody's single API approach for building AI applications over private data across multiple formats.
- [Diving Into AI with JavaScript Made Easy](https://www.youtube.com/watch?v=pjHDE3maXvE) — A live demo presentation at Eindhoven JS meetup covering getting started with AI in JavaScript, including data linking, semantic search, and generative search techniques.

## Organizations
- Unbody Lab (R&D lab) — An R&D lab building experimental software for making AI feel less artificial.
- Status (Creative technology studio) — A creative-technology studio building web products and design tooling across a group of branding companies.
- SUSLIB (R&D lab) — An R&D lab working on the future of knowledge interaction through computational design, research, and machine learning.
- ArtEZ University of the Arts (Art school) — A university of the arts in the Netherlands.
- Royal Academy of Art, The Hague (KABK) (Art school) — The Royal Academy of Art in The Hague.
- RNDR Studio (Creative technology studio) — A studio working in creative coding, data visualization, and interactive environments.
- LUSTlab (Design studio) — The interactive and technology arm of LUST, a graphic and interactive design studio.
- MindDesign (Design studio) — A design studio led by Niels Schrader, working between graphic design and physical media.
- MIT Media Lab (Research lab) — MIT's research lab for design and technology.
- Islamic Azad University, Najafabad (University) — A university in Isfahan, Iran.

## About
- About Amir: https://amir.cloud/about.md
- How Amir works — working style, strengths, collaboration: https://amir.cloud/profile.md
- Organizations, experience & education: https://amir.cloud/organizations.md