# Ahmad Shah — Product Engineer & Systems Architect

> Ahmad Shah is a product engineer specializing in full-stack platforms, clinical workflows, and AI systems with deep domain rigor.

- Website: https://iahmadshah.me/
- Blogs: https://iahmadshah.me/blogs/
- Design System: https://iahmadshah.me/design/
- Full Context Manifest: https://iahmadshah.me/llms-full.txt
- GitHub: https://github.com/ahmadshah2103
- LinkedIn: https://linkedin.com/in/ahmadshahkhattak

---

## Core Technical Focus
- **Full-Stack & Systems Engineering**: Astro, React, Next.js, Node.js, TypeScript, PostgreSQL.
- **AI & Retrieval Architecture**: Retrieval-Augmented Generation (RAG), pre-filtered vector search (PostgreSQL pgvector, ltree), LLM orchestration, streaming token pipelines.
- **Clinical & High-Stakes Domain**: Clinical decision-support systems, HIPAA/RBAC hierarchical authorization, real-time medical telemetry.

---

## Core Projects & Case Studies

### 1. Outpost — Natural Language AI Layer (Jarvis)
- URL: https://iahmadshah.me/work/outpost-ai-layer/
- Focus: Dynamic Supervisor/Anonymous-Worker multi-agent topology in LangGraph, least-privilege capability filtering across 300+ ops, human-in-the-loop write safety, live Mongoose company-roster caching, and model-metered AI billing.

### 2. Clinical Workflow Assistant
- URL: https://iahmadshah.me/work/clinical-workflow-assistant/
- Focus: Production HIPAA-aligned clinical documentation, LangGraph migration, XState encounter machines, and longitudinal patient memory models.

### 3. School Management System
- URL: https://iahmadshah.me/work/school-management-system/
- Focus: Full-stack lead, multi-role RBAC architecture, dynamic fee versioning engines, and zero-downtime PostgreSQL schema migrations.

### 4. Work Archives & Exploratory Systems
- URL: https://iahmadshah.me/work/archives/
- Includes:
  - **AI Receptionist Dashboard**: https://iahmadshah.me/work/ai-receptionist-dashboard/ (30-day serverless AWS MVP · Strict NDA)
  - **Promptis — Permission-Aware RAG Assistant**: https://iahmadshah.me/work/promptis/ (Pre-retrieval authorization over PostgreSQL/pgvector/ltree · Open Research)

---

## Architectural Writing & Articles

### 1. Stop Optimizing for Search Engines: How Agents Actually Read Your Site
- URL: https://iahmadshah.me/blogs/from-seo-to-aieo/
- Summary: Why client-hydration payloads and loose meta tags starve LLM context windows, and how to build deterministic machine readability.

### 2. Securing Multi-Tenant RAG: Recall Collapse, pgvector Filtering, and PostgreSQL ltree
- URL: https://iahmadshah.me/blogs/authorization-before-retrieval-in-rag/
- Summary: A deep dive into multi-tenant RAG authorization: why in-memory post-filtering causes recall collapse, how pgvector HNSW scans execute internally, and building hierarchical RBAC with PostgreSQL ltree.

### 3. The Fast Shipping Strategies That Quietly Kill Products
- URL: https://iahmadshah.me/blogs/product-strategy-failures/
- Summary: An analysis of counter-intuitive failure modes in fast-moving engineering teams and why velocity without architectural conviction destroys product defensibility.

### 4. Refactoring Is Often Delayed Design
- URL: https://iahmadshah.me/blogs/refactoring-is-delayed-design/
- Summary: Refactoring is often the compounding cost of assumptions that were never pressured early enough.

### 5. Questioning Features Before Users Have To
- URL: https://iahmadshah.me/blogs/questioning-features-before-users-have-to/
- Summary: How pressure-testing edge cases and user workflows before implementation prevents architectural drift.