AI, Bitcoin, Operations
San Francisco / Bombay
I'm genuinely drawn to emerging technology — I co-founded Vara Capital, a Bitcoin fund, and have worked with dozens of blue-chip crypto projects across exchange listings, operations, treasury, and governance. I also contribute to open-source AI agent tooling. More recently I've focused on bringing that same rigor to small businesses and independent operators as a fractional COO, building the systems and back-office that let them run properly without a full executive team.
My background spans an MBA, economics, and early career work in structured risk — including asset protection planning and specialized trust development — which still shapes how I think about structure and downside today.
Fractional COO support for independent businesses and solo practitioners — cleaning up the admin, processes, and systems, building accountability structures, and layering in AI tooling so the business runs with significantly less manual overhead.
Helping individuals and businesses hold Bitcoin properly and compliantly — multisig wallet setup, recovery and inheritance planning, and long-term allocation strategy for people who want to do this right, not just buy and hope.
Account management and operations for crypto-native organizations — coordinating compliance, treasury, and stakeholder relationships across multisig governance and cross-jurisdiction structures.
Websites, open-source tooling, and AI tools I've designed and shipped end to end.
Brand and site for a Bitcoin momentum fund — performance metrics, fund structure, and investor information built into a clean, credible site.
Brand and website for a functional nutrition and health coaching practice — positioning, visual identity, and a site built to make booking easy.
An AI agent governance platform — site and product built solo, end to end.
A career-risk tool that gives a straight read on when AI is likely to affect a given job, and what to do about it.
Continuous 1–5 risk scoring for AI agent actions. An open-source alternative to binary allow/deny authorization.
View repository →An MCP server that sits between AI agents and real systems, evaluating every tool call before execution and holding risky ones for review.
View repository →Streaming pattern detection for AI agent actions across systems — catching multi-step risks single-action authorizers miss.
View repository →A local, model-agnostic agent that reads email purchase receipts and builds a structured wardrobe inventory — then reasons about gaps, color balance, and what to buy next. Runs entirely on your own machine; no data leaves it.
View repository →A custom GPT that forces a real decision out of you instead of more options. Tell it what you're choosing between, why you're stuck, and what actually constrains you — money, time, risk, the people involved — and it pushes you to a call.