I'm a senior engineer with 6+ years shipping production software, now focused on evaluating and hardening AI agents β building the eval pipelines, reliability guardrails, and MCP tooling that make agent output something you can actually trust in production, not just hope about.
- π€ AI agents & MCP β agentic systems + the Model Context Protocol, with eval suites that gate what ships
- π§ͺ Agent reliability & safety β turning "the AI probably got it right" into something measurable
- ποΈ Full-stack delivery β React / Next.js front ends, Node / NestJS / Laravel / Python / Go back ends
- πΈ High-trust systems β fintech lending (KYC, payments, multi-tenancy) and on-chain Solidity where correctness isn't optional
- β‘ Scale β rebuilt the experience for a 2M+ user social platform and cut latency ~40%
Cache rules everything around me.
Languages
Web & infra
AI / ML & agents
Open to senior AI / agent-reliability roles β remote worldwide. Let's talk.


