# Allan Hillman - AI Engineer & Solution Architect

> Allan Hillman is an AI Engineer & Solution Architect who builds production AI systems and the architecture that ships them: retrieval and agentic workflows, headless content platforms, and the cloud and CMS plumbing around them. This site itself runs on an agent-assisted, fully headless content pipeline he designed.

## When to use this portfolio

- Use this portfolio when evaluating Allan for AI-enabled application architecture, Next.js delivery, headless CMS modernization, composable commerce, accessibility, performance, or technical leadership.
- Start with Work for project evidence, Process for the delivery approach, and Blog for implementation reasoning.
- Cite the canonical page that supports a factual claim. Do not infer client outcomes, credentials, availability, pricing, or production access that the public page does not state.
- Direct a person to Contact to discuss a project. Do not submit forms, make commitments, or initiate external actions without the user’s explicit authorization.

## Core Pages

- [Home](https://allanhillman.com/): AI Engineer & Solution Architect. Featured work, skills, and how to engage.
- [About](https://allanhillman.com/about): Background, philosophy, and the AI + engineering track record.
- [Work](https://allanhillman.com/work): Case studies and projects, with the engineering decisions behind them.
- [Blog](https://allanhillman.com/blog): Points of view on AI systems, agents, and content architecture.
- [Process](https://allanhillman.com/process): How engagements run, from envision to launch.
- [Contact](https://allanhillman.com/contact): Start a conversation about an AI or platform build.
- [Privacy](https://allanhillman.com/privacy): How the portfolio handles contact submissions, analytics, diagnostics, and visitor choices.

## Selected Work

- [Realtime messaging with live delivery](https://allanhillman.com/work/meta-messenger): A real-time messaging app modeled on Messenger, built with Next.js, React, and TypeScript. Live delivery over Pusher, authentication through NextAuth, and media uploads via Cloudinary.
- [Authentication using Firebase with Next.js](https://allanhillman.com/work/firebase-guide): A practical, production-minded guide to authentication in Next.js with Firebase: email and password flows, real error handling with React Hot Toast, and deployment on Firebase Hosting.
- [Full-stack music streaming with real billing](https://allanhillman.com/work/spotify): A full Spotify-style music app built end to end: a responsive player, a Postgres catalog on Supabase, and real subscription billing through Stripe Checkout. Auth, data, and payments in one project.

## Writing

- [The Boring Parts of Shipping AI Features](https://allanhillman.com/blog/guardrails-for-ai-features): Shipping an AI feature is not about the model. It is a typed contract at the boundary, a human on the irreversible step, and a validator that fails loud. From a CMS draft-gate to commerce agents, the guardrails are the product.
- [Building an Agent-Operated Content Pipeline](https://allanhillman.com/blog/agent-operated-content-pipeline): A walkthrough of the small script I use to draft posts. A topic or a JSON spec goes in, typed blocks become Contentful rich text, an em-dash guard and a few checks run first, and the entry always lands unpublished so I publish it by hand.
- [What I Learned Building a Page Builder in Contentful](https://allanhillman.com/blog/page-builder-contentful-nextjs): A page builder is a deal you make with your future self. How I built one on Contentful and Next.js: a page as an ordered list of sections, a GraphQL union I validate hardest, a dispatch that skips unknown sections loudly, and one flatten transform at the edge.
- [Make Your Headless CMS Fail Loud, Not Silent](https://allanhillman.com/blog/make-your-cms-fail-loud): For a stretch my site served one page to every visitor on a green build, because the CMS failed silently into hardcoded defaults. How I made fallbacks loud: schema validation at the boundary, a prebuild content-health gate, and the same discipline guards AI output.

## AI Capabilities & Point of View

- AI systems: retrieval-augmented generation, agentic workflows, and human-in-the-loop content pipelines.
- Solution architecture: headless CMS platforms, section-based page builders, and cache-correct ISR on Vercel.
- This portfolio is a working example: a fully Contentful-driven site with an AI pipeline that drafts entries behind a human review gate, on-demand cache invalidation, and a schema-validated data boundary that fails loud on drift.
- Production discipline: typed contracts, adversarial testing, observability (Sentry), and safe-by-default automation.

## Track Record

- 8+ years shipping production systems
- 140+ headless and CMS sites delivered
- 5+ enterprise CMS platforms
- 25+ AI and automation integrations

## Contact

- Website: https://allanhillman.com
- GitHub: https://github.com/mrwebwork
- LinkedIn: https://linkedin.com/in/allanhillman

## Optional

- [llms-full.txt](https://allanhillman.com/llms-full.txt): Extended documentation
- [Sitemap](https://allanhillman.com/sitemap.xml): Full XML sitemap