How to Optimize Your Portfolio Website for AI Search (2026)

Q: How do I measure AI visibility for my portfolio?

Monthly audit: ask ChatGPT/Perplexity 'who is [your name]?' and 'what is [your project]?' and note whether your site is cited; watch AI-referrer traffic in analytics; and track whether branded searches grow. There's no dashboard — build the audit habit.

Layer 1: The Identity (Person Schema)

AI engines need to resolve "who are you" before they can cite you. The Person block with sameAs links does that: your name, site, GitHub, and profiles become one entity. The template is in the JSON-LD cheat sheet — the key fields:

name, url, jobTitle — the basics, unambiguous.
sameAs — GitHub, LinkedIn, and any other verified profiles. This is what lets an AI assistant connect "Hussain Nazary the portfolio author" with "Hussain Nazary the GitHub contributor."
knowsAbout — your skill list in machine-readable form (RAG, local LLMs, vector databases…).

Layer 2: Projects That Get Cited

Project descriptions are usually teasers. AI engines need answers. The structure that gets quoted:

First sentence = the answer: "[Project] is a privacy-first legal assistant that answers questions over confidential documents using a local RAG pipeline." Done — that sentence is the citation.
Problem → Solution → Stack → Outcome, explicitly labeled. "Problem: … Solution: … Stack: BGE-M3, ChromaDB, Ollama. Outcome: …" — extraction-friendly structure.
Measurable outcomes: "cut document review time by 60%" beats "improved efficiency".
SoftwareApplication/CreativeWork schema on project pages, linked to your Person block.

Layer 3: The Blog Is the Citation Surface

This is the layer that compounds. A portfolio blog that answers real questions — "what is GGUF?", "how to build RAG?", "best model for 16GB RAM" — creates hundreds of pages that AI engines can quote. Every AEO principle applies per post:

Direct answer in the first 50–100 words
Question-matching H2s and FAQ sections with FAQPage schema
Data, tables, and sources (citable specificity)
Consistent structure so extraction is reliable

Each post also links back to the portfolio projects — which is how blog citations become project citations. The blog here at Hussain-Nazary.github.io is built exactly this way: 40+ posts, every one with BlogPosting + FAQPage JSON-LD, direct-answer openings, and internal links to the flagship projects.

Layer 4: The Technical Floor

Check	Why
sitemap.xml	The crawl map — every page listed, dated, prioritized. Validate it.
Unique meta descriptions ≤160 chars	What engines and AI systems display; duplicates get ignored
One H1, clean heading hierarchy	The extraction skeleton — question H2s are citation targets
Canonical tags	One URL per page; prevents identity splitting
No JS-hidden content	If the answer only renders after JS, crawlers may never read it
HTTPS + fast load	Baseline trust and crawl budget

The Monthly AI Audit

Ask three engines: "who is [your name]?", "what is [your flagship project]?", "what does [your name] do?"
Record: cited? cited correctly? linked to the right pages?
Check analytics for AI-referrer traffic (chatgpt.com, perplexity.ai, etc.).
Fix what's wrong: wrong attribution → tighten sameAs; no citation → strengthen the direct answer on the relevant page.

🚀 The compounding loop

Every blog post is a new citation surface; every citation trains the engines that you're a source; every source-citation makes the next answer more likely to include you. It's slow at first and fast later — which is exactly why starting now (and publishing consistently, as this site does) is the strategy.

Frequently Asked Questions (FAQ)

How do I make AI assistants find my portfolio?

Three layers: (1) a Person JSON-LD block with sameAs links so your identity is machine-resolvable; (2) project pages written with direct-answer openings and clear structure so they get quoted; (3) a blog of question-answering content that becomes your citation surface. Plus a sitemap and working meta descriptions.

What is the most important schema for a portfolio?

Person — it ties your name, site, GitHub, and other profiles into one machine-readable identity via sameAs. Article/BlogPosting for posts, and CreativeWork or SoftwareApplication for projects complete the picture. The JSON-LD cheat sheet in this blog has copy-paste templates.

How should I write project descriptions for AI search?

Like an answer, not a teaser: name the project, state what it does in the first sentence, list the problem/solution/tech stack explicitly, and include measurable outcomes. AI engines quote the clearest passage — give them a clean one to quote.

Does a portfolio blog actually help?

Dramatically. A blog gives you hundreds of pages that answer real questions ("what is GGUF?", "how to build RAG?") — and every FAQ section and direct answer is a citation target. The blog is how a personal site becomes a source, not just a destination.

What are the technical basics I must not skip?

A valid sitemap.xml, unique meta descriptions under ~160 characters, one H1 per page, working canonical tags, HTTPS, and no JavaScript-rendered content hiding your text. These are the crawlability floor — everything above is polish.

How do I measure AI visibility for my portfolio?

Monthly audit: ask ChatGPT/Perplexity "who is [your name]?" and "what is [your project]?" and note whether your site is cited; watch AI-referrer traffic in analytics; and track whether branded searches grow. There's no dashboard — build the audit habit.