Writing & Insights

Blog

Technical deep dives, product thinking, and stories from building safety-critical systems at scale.

AI Safety

March 16, 2026·12 min read

1,405 Ways to Break an LLM

Analysis of 1,405 LLM jailbreak prompts across 11 attack families - from DAN roleplay to indirect prompt injection. Covers jailbreak techniques, defense-in-depth, CaMeL, and step-by-step guidance for securing agentic AI systems.

Read article

Autonomous Driving

Product ThinkingAutonomous Driving

March 11, 2026·9 min read

NVIDIA's Alpamayo: When Reasoning Meets the Road

A visual explainer of NVIDIA's Alpamayo — how a 10B-parameter model reasons through driving decisions in 99ms, with interactive diagrams breaking down the architecture.

Read article

AI Safety

Product ThinkingAI Safety

February 9, 2026·11 min read

Amodei Wrote the Risk Assessment. Who's Writing the Runbook?

A practitioner's critique of Anthropic CEO Dario Amodei's "The Adolescence of Technology" — what operationalizing AI safety actually requires, from someone who built safety-by-design at Amazon scale.

Read article