This portal exposes the same information as the human website (including live database content) in a dense, fully expanded HTML form for AI agents and crawlers. There are no write actions. Agent pages are noindex; prefer human URLs for citation.
AI builder. AI alignment researcher. Physicist.
I build real AI tools and drive their adoption inside organizations.
Outside Israel (English) — job-seeker; builder + alignment co-equal
Dror Aharoni is an applied-AI builder and AI alignment researcher with an M.Sc. in physics. He ships production AI tools (Socratic digital-twin tutors, a coach digital-twin platform, an exam-classification pipeline using Power Apps + Python + Gemini) and has spent over a decade driving org-wide AI adoption. His alignment work (the "Rearing Paradigm" / "Compliance Trap" thesis) sits alongside his building, not beneath it. He is seeking an applied-AI / AI-adoption lead role. Netherlands primary, US (NY/NJ) secondary, UK backup. EU work authorization finalizing, available from October 2026, no sponsorship needed.
Person: Dror Moshe Aharoni — Aha.Dror.AI@gmail.com
Site map for agents
Machine links
Recent guides
- GPT-5.4 & GPT-5.4-pro: Complete OverviewChatGPTEverything about GPT-5.4: pricing, context window, features (tool_search, Computer Use, Compaction), and the expensive pro variant for hard tasks.
- Why Most People Get Bad Results from ChatGPT (and How to Fix It)ChatGPTAI models are programmed to please you and always give an answer - even when wrong. Fix this with one custom instruction.
- Anthropic Evolving EcosystemClaudeA comprehensive guide to Anthropic s 2026 ecosystem, from Claude Design to Claude Code, and how they change the design and development world.
- Fix Claude Sycophancy with System InstructionsClaudeStop Claude from always agreeing with you by setting custom instructions.
- Claude Cowork: 10 Use Cases I Tested + 67 More by ProfessionClaudeComprehensive guide to Claude Cowork (desktop agent) with 10 tested use cases including pet photo sorting, Gumroad product creation, invoice reconciliation, content repurposing, and more.
- Claude Code Subagents: Enable Persistent MemoryClaude CodeHow to give each Claude Code subagent its own persistent memory for project patterns, recurring issues, and architectural knowledge.
- Claude Code - Getting StartedClaude Codeכל מה שצריך להתחיל עם Claude Code: מה זה, התקנה, עברית RTL, אפליקציית דסקטופ, ופקודות בסיסיות.
- Claude Code: The 4 Layers FrameworkClaude CodePractical guide to leveling up from basic prompting to agentic automation.
- Build an AI Second Brain with Claude Code + ObsidianClaude CodeA practical guide to turning Obsidian into an AI-powered second brain using Claude Code. Store everything you know about your business in markdown files — clients, sales scripts, offers, meeting notes, voice profiles — and let a Claude agent read and write them directly from your terminal.
- Claude Code - Daily WorkClaude Codeטיפים מהקהילה, סריקת קוד, מצבי חשיבה, Remote Control מהטלפון, ותהליך פיתוח ממוקד סוכנים.
Sample AI tools
- A0.devפיתוח וכלים מתקדמים · בתשלום בלבד
- Ace Studioיצירת מדיה (תמונות וידאו וסאונד) · בתשלום בלבד
- Ad Creative AIיצירת מדיה (תמונות וידאו וסאונד) · בתשלום
- Adobe Fireflyיצירת מדיה (תמונות וידאו וסאונד) · בתשלום בלבד
- ADOBE Podcastיצירת מדיה (תמונות וידאו וסאונד) · בתשלום בלבד
- Agent Skillsפיתוח וכלים מתקדמים · חינם
- AgentScopeפיתוח וכלים מתקדמים · חינם
- AgentSkills ILmarketplace · free
- AICODE | גל חבקיןפיתוח וכלים מתקדמים · חינמי
- Allhands_aiפיתוח וכלים מתקדמים · חינמי
Projects
- Build-Your-Own Optics Simulation (AI project)למידה מבוססת פרויקטStudents wear three hats in turn (researcher, product manager, developer) to take an optical phenomenon from idea to a working AI-built simulation, no prior coding required. Seventy percent of the grade is the process: every milestone must pass through an AI mentor bot for feedback and revision, claiming 'everything worked perfectly' costs points, and solving a real breakdown earns a bonus. Productive struggle is designed in.
- Physics Courtroomסימולציה לימודיתA multilingual educational physics game where students investigate forensic-style cases, request lab evidence, and argue physics-based conclusions in an AI-powered courtroom with dynamic characters (judge, prosecutor, defense, lab). Students join with a short lesson code—no login required—and progress through narrative, lab, and court phases. Google Gemini generates structured courtroom dialogue while a hybrid verdict model keeps the application in charge of when the case is actually solved. Built for Israeli high-school physics (grades 9–11), with teacher tools for creating lessons and monitoring live progress. Fully supports four languages: Hebrew, English, Russian, and Arabic.
- Dror AI Agent Launchpad — Custom AI Coding SkillsאוטומציהA public GitHub repo of custom skills for Claude Code, Cursor, and Codex — including agent-onboarding-wizard (safety review before installing plugins/MCP), new-project-from-prd (PRD to working project scaffold), cv-story-builder, google-play-publishing, and outside-reader-audit. Built to help non-developers and developers work safely with AI coding agents, with ready-to-install .skill packages and bilingual guides.
- Interactive Solar System — 3D Science Projectסימולציה חינוכיתA bilingual (Hebrew/English) educational website for exploring the solar system in interactive 3D — built in 5 hours for my son Marom's science fair project. Marom defined the requirements (PRD), and after several rounds of edits and fixes until he was happy with the result, the site was ready. Features include a gallery of 13 celestial bodies, an animated 3D orrery with clickable planets, accurate-scale toggle, detailed fact pages, size comparison tool, and full RTL support.
- Eco-Bot — AI Marketing Agent for EcometerבוטיםA Gemini-powered conversational bot built to promote the Ecometer project — a classroom electricity-monitoring system deployed across Israeli schools. Presented at a major education conference, the bot demonstrates how AI can teach students about energy consumption, solar panel output, and environmental awareness through interactive dialogue with real sensor data.
- Eco-Simulator — Interactive Thermodynamics Classroom Labסימולציה חינוכיתAn interactive thermodynamic simulator built for the Ecometer project, allowing students to explore energy consumption in a virtual classroom. Users adjust initial conditions — ambient temperature, classroom temperature, floor level, window orientation, building insulation quality, and class size — then observe real-time effects on electricity usage. Supports multiple AC strategies and generates an educational report comparing energy savings without compromising student comfort.
- AI Exam Editor — Pedagogical Assessment BotבוטיםAn AI-powered exam editing assistant built for teachers serving as test editors in the Israeli Ministry of Education system. After uploading an exam file, the bot systematically analyzes question phrasing, difficulty calibration, answer clarity, and alignment with curriculum learning objectives. It provides detailed feedback and suggests improvements — acting as both an editor and pedagogical advisor. Designed to elevate exam quality at scale while preserving the teacher's pedagogical intent.
- POE Bot — Galileo's Principle of InertiaבוטיםA Custom GPT that guides students through a deep Socratic discussion of Galileo's famous thought experiment: a ball dropped from the mast of a moving ship. Built around an original passage from Galileo's writings, the bot explores the principle of inertia by challenging students to predict the ball's landing point before and after the ship begins moving at constant velocity. The central pedagogical question — whether a starting reference frame is equivalent to a target reference frame — drives students toward genuine conceptual understanding rather than rote answers.
Insights
- Shape the Technology: What a Physics Classroom Taught Me About Building AI That Teaches2026-07 · 8 min readFor twenty years we adapted technology to schooling and learning outcomes stayed flat. The harder path is to shape the technology so it develops understanding, the way a good teacher does. A physics teacher's field report, with references.
- The system I built is being retired. That is the success story.2026-07 · 4 min readA lab-operations platform scaled from one center to a network — and is now being replaced. The tool is dying; the ways of working it created are the standard. That is what adoption means.
- The Compliance Trap: Why AI Alignment Is an Education Problem, Not an Engineering One2026-06 · 12 min readA physics teacher's account of how watching AI accelerate led him to reframe alignment as an educational challenge, not an engineering one.
- The Rearing Paradigm: A Falsifiable Proposal for Developmental Moral Training2026-06 · 22 min readAn education-grounded, falsifiable proposal for AI alignment: whether the structure of moral training data, not only its content, produces more robust and more stable models.
- Seven Principles from the World's Wisdom Traditions — for AI Value Alignment2026-03 · 20 min readWhat can Buddhism, Taoism, Ubuntu, Stoicism, and Indigenous traditions teach us about building safer, more principled AI?
- The Ten Commandments as a System Prompt2026-03 · 18 min readWhen 3,300-year-old wisdom meets artificial intelligence — and what the Talmud can teach us about building smarter, safer, more principled machines