I design and implement governed AI systems and workflows, with a focus on agent reliability, eval design, behavior and control specifications, authority boundaries, review gates, and inspectable execution.
Second Mind Systems works across AI systems architecture, workflow diagnosis and implementation, agent reliability, evaluation, behavioral controls, and governed AI workflows.
Open to consulting, implementation work, strategic collaborations, and full-time roles in AI systems, agent reliability, evals, workflows, product strategy, and governance.
Contact: secondmindsystems@gmail.com
- Behavior Profiles — portable conduct contracts for making expected AI-agent behavior and boundaries explicit through durable instruction surfaces. Persistence means the profile remains available to the agent; it does not mean universal obedience or enforcement.
- Governed Change Demo — a deterministic repository-change demonstration that evaluates declared authority, path scope, claims, and evidence before returning an acceptance result, with inspectable receipts. Run it in five minutes.
Bring one recurring AI behavior problem in one bounded workflow, with real examples.
For every qualifying Lab engagement, you receive a written specification of the required behavior—what the agent should do, what it must not do, and when it should stop, ask, or escalate—together with relevant boundaries and authority conditions, observable criteria for testing the behavior, application guidance, and explicit limitations.
Where the evidence supports a formal Behavior Profile, that portable behavior artifact is delivered. Where it does not, the engagement returns the intervention or conclusion the evidence supports rather than forcing a Behavior Profile.
The Lab is a complete bounded control-design engagement. Production code changes, integrations, deployment, live-system operation, and tests executed against your environment are separate execution surfaces unless explicitly scoped.
Founding offer: $200 for one qualifying bounded engagement. First 10. Delivery is within seven calendar days once payment and the required intake/material are complete.
For fit intake, email secondmindsystems@gmail.com with the subject Behavior Profile Lab — Fit Intake and a short description of what keeps happening. Please do not send credentials, private code, sensitive records, or other confidential material in the first message.
I work hands-on across problem framing, architecture, implementation, evaluation design, authority boundaries, acceptance criteria, and final technical review. AI agents may assist with bounded implementation and review. I remain responsible for defining the work, constraining what may change, checking the result against repository and evidence truth, and deciding what is accepted.
The linked public proof demonstrates bounded artifacts, documented methods, deterministic evaluation mechanisms, and reproduced local results within the scopes stated by those repositories. It does not establish universal agent conformance, cryptographic security, compliance certification, production deployment, customer validation, market validation, or access to private Second Mind runtime machinery. Each public repository defines its own evidence and limitations.
View Behavior Profiles · Run the Governed Change Demo · Contact Second Mind Systems