Engineering systems and the evidence behind them

Evaluation, applied AI, and agent workflows

Useful automation starts with a defined task and a way to examine what the system actually does. I build agent workflows and the instrumentation needed to compare alternatives, identify failures, and support human decisions about responsibility.

Applied AI engineering

Agent loops, explicit pipelines, tool and model integrations, and operational workflows with inspectable outputs.

Evaluation and failure analysis

Task definitions, grading rubrics, evaluation harnesses, tracing, human feedback, offline replay, and paired experiments.

Workflow architecture and human review

Durable execution, artifact handoffs, operator views, and review policies that make asynchronous work easier to inspect and direct.

Payment infrastructure

I design and implement payment flows using x402, smart contracts, and blockchain rails. Work can include API billing, micropayments, agent-to-agent transactions, and controls around consequential transfers.

Protocols and integration

x402 payment flows, smart contract development, Base and Ethereum L2s, and USDC or other stablecoin infrastructure.

System design

Payment authorization, transaction verification, and integration with the surrounding agent workflow.

Blockchain and zero-knowledge cryptography

I work on smart contracts, zero-knowledge circuits, and privacy-preserving systems. The approach starts with the property a system needs to establish and the assumptions under which a proof or protocol supports that property.

Implementation

Solidity smart contracts, Noir and Circom circuits, Semaphore-based privacy systems, and on-chain coordination.

Architecture and integration

Verifiable computation, cryptographic audit trails, private state, and integration with software and agent systems.

Discuss your system

Tell me what you're building, what needs to work, and where you need better evidence.