Years of experience building technical solutions across unconventional settings — from engineering FedRAMP-compliant ML systems for the national COVID response to deploying agentic AI for survivors navigating hostile institutions. I don’t lead with a technology. I start with your problem, then find the simplest solution that actually works.
When AI helps, I use it — specifically Claude, which I deploy through Claude Code as a development collaborator. Judgment + execution: I decide what to build, in what order, and to what standard. Claude executes with tireless coherence and context-holding. Neither half is optional.
I don’t treat Claude’s output as ready-to-ship. Every implementation gets review, feedback, and iteration. Some of Claude’s suggestions are wrong — plausible-sounding but incorrect. That’s where I add value: knowing what to trust, what to verify, and where to push back.
The result: infrastructure that would normally require a 2–3 person engineering team, delivered in weeks. You get someone who understands both your problem and the tool — and who will tell you honestly when AI isn’t the right answer.
Survivor & Advocate Applications
Agentic AI has specific value for people in difficult institutional situations — survivors of abuse, activists, whistleblowers, people navigating coercive control.
- Evidence & documentation: Automated workflows that create timestamped, organized records of abuse, harassment, or institutional misconduct. The documentation is encrypted, tamper-resistant, and available in legal proceedings.
- Institutional navigation: Legal documents, FOIA requests, formal complaints — Claude can help parse the language, understand what you’re reading, and draft communications that are precise and compelling.
- Records management: Victims of coercive control often have fragmented digital lives. Recovering and organizing what exists, and creating a secure new structure, is essential work.
The Threat Model
AI compresses the timeline between vulnerability discovery and exploitation. If your adversary has technical access or AI literacy, that asymmetry is the central design constraint. Every survivor application I build starts from this: your adversary may be using the same tools. We build accordingly.
Engagement Models
Hourly for sessions, audits, and implementation work. Project-based for discrete deliverables with clear scope. Retained monthly for ongoing technical partnership.
Nonprofits, students, and survivors can ask for a reduced rate on this work. It slides down toward $100 an hour and no further, because $100 is the floor across the whole practice. Ask directly and I will give you a number.