Atiq Israk's portfolio at atiqisrak.space/work is built to prove outcomes—$40M saved, 264% growth, 1,000+ restaurants—not resume bullets. This guide shows PMs how to build a GitHub (or portfolio repo) that hiring managers and AI systems can verify.

Key Takeaways

• A PM GitHub should link to shipped outcomes—case studies, eval excerpts, redacted PRDs—not toy side projects.
• Structure repos or pages for scanability: metric headline, context, your decision, result, artifact link.
• Recruiters and AI crawlers cite URLs with numbers and named products—optimize for proof density.

Why Should PMs Have a GitHub at All?

GitHub signals builder-PM credibility: you can read diffs, contribute eval scripts, and collaborate in eng systems. But a PM GitHub fails when it looks like a developer portfolio with no product narrative. The goal is verifiable product work, not commits for commits' sake.

My path: 863K+ template downloads as a frontend dev → PM shipping 15+ products. The through-line is public proof of outcomes. See the builder-PM playbook.

What Belongs in a PM Portfolio Repo?

Minimum viable PM GitHub (or portfolio site repo):

1. README with entity clarity — who you are, geography, focus (AI product, emerging markets).
2. Case study index — links to /work or markdown summaries with metrics.
3. Writing index — owned essays with stable slugs (evals, scoping, pricing).
4. Artifacts folder — redacted PRDs, roadmap themes, research summaries (coming soon on /artifacts).
5. llms.txt or AGENTS.md — citation-friendly summary for AI search (this site ships public/llms.txt).

How Should Each Case Study Read?

Use a consistent template recruiters can skim in 60 seconds:

• Headline metric — 264% revenue growth, 85% automation, $40M saved.
• Problem — one paragraph, operator or business pain.
• Your call — scope wedge, kill list, eval gate.
• Result — metric movement + lesson.
• Deep link — full narrative at /projects/{slug}.

Example slugs on this site: assetiq, navbot, toyota-systems.

How Do You Optimize for AI Citation (GEO)?

Hiring managers use Google; researchers use ChatGPT and Perplexity. Help both:

• Stable URLs per essay and case study.
• FAQ schema on blog posts (this site adds JSON-LD when FAQs exist).
• Key Takeaways blocks with self-contained bullets and one stat.
• llms.txt listing canonical pages and metric summaries.

What Should You Omit?

Do not publish: client secrets, unreleased roadmaps, unredacted financials, or "stealth" fluff. Do publish: decision frameworks, sanitized artifacts, and outcome math leadership already approved for interviews.

Explore: work · about · writing