AI agents that do the work, within a budget you set.

Agents are easy to demo and hard to trust. I build agentic workflows and multi-agent systems for startups with the controls that make them safe to run unattended: clear tools, guardrails, retries, spend limits and full traces of what each agent did.

What I build

Single agents that call your APIs and tools, multi-step workflows that research, draft and act, and multi-agent systems where many agents simulate or divide work. Each one runs on a durable backend with queues and retries, not a chat loop that dies on the first timeout.

Common questions

What does an AI agent developer do?

An AI agent developer builds LLM systems that take actions, not just answer questions: calling tools and APIs, planning multi-step tasks and coordinating with other agents, with guardrails so they behave predictably in production.

How do you stop agents from running up a huge bill?

Every agent system I build has model routing, per-run spend ceilings and a watchdog that cancels runs past the cap, plus dashboards for token spend per task.

Which frameworks do you use for agents?

I pick per project, from plain code with an LLM SDK to frameworks like CrewAI or Agno. Durable execution, observability and cost control matter more than the framework.

Shipped work

  • Swarmie

    500-agent LLM swarm that validates a startup pitch in about 60 seconds, with a hard cost cap.

  • Pitchcraft

    LLM pipeline that audits sites, writes briefs and deploys a prototype per lead, with multi-provider fallback.

  • Socella

    Founding AI engineer. Vector search, agentic automation, RBAC and Stripe. $15K+ early revenue, $50K+ raised.

Reply within 24 hours