Now
What I'm focused on, right now.
A short snapshot, updated when something meaningful changes. Inspired by the /now movement.
Last updated April 28, 2026Building
- Project Netra — shift-left controls for AI coding assistants at Adobe
- Adobe Cyber Trail (ACT) — turning fragmented security signals into automated response
- RAG + multi-agent experiments on the side (LangChain / LangGraph)
Writing
- An essay on the secure-blocks methodology
- A short series on LLM-assisted threat modeling
- Weekly notes on what's actually working with AI in production security
Speaking & advising
- M3AAWG, BSidesSF, OWASP SF chapter sessions in 2025
- Advisor — Strategic AI Program, USF School of Management
- Active member, Cyberpeace Institute
Testing
- Local model setups for offline drafting and triage
- Long-running research agents for slow background work
- Prompt-as-API patterns — versioned, observable, owned
Reading & links
- The Bitter LessonReading
Re-read every quarter. The compass for AI work. — Rich Sutton
- Software 3.0Reading
Models as the new abstraction layer — clean framing. — Andrej Karpathy
- Prompts as APIsReading
Treat prompts the way you treat endpoints: versioned, observable, owned. — Personal note
- Deep Research LoopsTools
Why a slow agent that's right beats a fast agent that's confident. — Build log