About
I'm a software engineer with 9+ years across distributed systems and Web3: JPMorgan, Crypto.com, Gelato, and currently Moralis, where I build production agentic-AI infrastructure. That means a Grafana MCP server, code-review agents that remember what they've seen before, an automated reviewer that goes over a work session after it ends, and eval harnesses for RAG systems. It's the boring plumbing that decides whether an "AI feature" actually holds up once real traffic hits it.
Outside work I ship open-source Claude Code plugins, ContextSpin and RecallCheck, and keep a running habit of publishing real, measured experiments instead of tutorials or predictions. I'd rather post one number I actually observed than ten opinions about where the field is going.
Background
- 9+ years, distributed systems & Web3
- Currently: Senior Software Engineer @ Moralis
- Past: JPMorgan, Crypto.com, Gelato
- Ships: Claude Code plugins (ContextSpin, RecallCheck), production MCP/agent tooling
I'm open to a small number of consulting conversations with teams working on similar problems: agentic tooling, evals, the parts of this that only break in production. Get in touch.
I'm also collecting interest for a small, private group for people building seriously with AI to share real experiments, no judgment unless you ask for it. If that's you, email me one thing you've tried and measured, and what happened.