Rohit’s AI Architect Playbook
I built this because I needed a disciplined way to stay sharp in AI architecture.
Not “AI news.” Not hype. Just a repeatable system: notice signals, think in systems, and write down decisions like they’ll be read six months later.
This is written for two kinds of readers:
- People getting started: plain-English explanations with small examples and exercises.
- People already building: architecture tradeoffs, reliability, security, and evaluation.
If you’re new, start with Daily. If you’re designing systems, jump to Patterns and Decision Records.
What you’ll find here¶
Daily
Short entries that force signal-awareness + architectural thinking. Two GenAI concepts, two agent concepts, one “AI radar” topic, and a design drill.
Executive Strategy
Weekly notes written like a Director thinking aloud — what caught my attention, what it might mean for insurance delivery, and where I’d be cautious.
Patterns
Reusable blueprints: retrieval, orchestration, observability/evals, guardrails, and failure handling.
Decision Records
“Why we chose this” docs: tradeoffs, risks, mitigations, and what we’d change next time.
If you want to collaborate¶
If you found something useful (or you disagree with it), that’s a good sign.
- If you want to track updates, you can star the repository on GitHub
- Open an issue with a counterpoint or improvement idea
- Fork the structure and adapt it for your own journal
No pressure — I’m optimizing for consistency, not popularity.