Researchingwhat matters.Building what works.
I'm Zawwar Sami — an independent researcher and engineer working across philosophy, artificial intelligence, and cybersecurity. I publish research, build systems, and test ideas beyond the page.
I build with ,
engineer with ,
and ship with .
I work independently across philosophy, artificial intelligence, and cybersecurity. Some questions become papers; others become systems I can actually test.
ZAI (Zawwar Autonomous Intelligence) and ZAI Memory Hub are part of that work. Anteroom Studio gives my research and engineering projects a home, from personal AI and open-source infrastructure to security and market tools.
Independent Inquiry
I follow questions across philosophy, AI, and technology without forcing them into one discipline.
Evidence Matters
An argument needs sources; a technical claim needs tests. I try to keep both open to scrutiny.
Security & Trust
Understanding systems means understanding where they fail, what they expose, and where their boundaries sit.
Build & Revisit
I publish, build, test, and revise. A good idea should survive contact with reality.
- Python
- PostgreSQL
- Next.js
- React
- TypeScript
- Node.js
- Tailwind CSS
- Supabase
- Docker
- GitHub
A focused, iterative process
- 01
Question
Start with the premise. What is being asked, and what would count as an answer?
- 02
Research
Read the literature, inspect evidence, and separate what is known from what is proposed.
- 03
Build & Test
Write, prototype, or investigate in a controlled setting. Find the limits of the idea.
- 04
Publish & Revise
Document results, show the sources and limitations, then improve when the evidence changes.
Systems I've built
- 01Open Source2026
ZAI Memory Hub — A shared memory layer for one human and many AIs
A Postgres-backed shared memory and decision layer for assistants working across sessions and tools. MCP exposes the records; the dashboard makes context, provenance, and revisions inspectable.
- 02Open Source2026
zhub — WiFi for AIs — make any AI a discoverable endpoint
A drop-in skill that lets any AI publish a discoverable, controllable, OpenAI-compatible endpoint in three commands. Bidirectional: the AI sees its connected clients and their capabilities; clients see what the AI can do. The intent was to turn 'connecting two AIs together' from a research project into something that looks like joining a WiFi network.
- 03
AI2026Anteroom Oracle — An AI terminal for geopolitical and macro intelligence
An AI-powered intelligence terminal that compresses geopolitical and macro signal into something a single operator can actually act on — with crisis replay, scenario simulation, and market-regime detection built in.
A question worth exploring.
Open to thoughtful correspondence, research exchange, authorized security work, and engineering collaborations where the problem is worth investigating.