Eric NolanUnyielding Claw
Technical Product ManagerHands-on AI DevelopmentAI Governance

I don't just roadmap AI.

I build the systems that make AI useful.

I design and build AI-assisted products, developer tools, context systems, governed automation, and the integration layer around them—then make the failure modes visible enough to trust in real work.

See what I've built How I engineer AI
PythonFastAPILangGraphLocal LLMsAgent orchestrationEvaluationGuardrails
AI DEVELOPMENT // EVIDENCE BUILDING
01
Context & Memory Systems

Authority, freshness, provenance, drift detection, hooks, and context ceilings.

02
Agent & Tool Orchestration

Specialist agents, local inference, APIs, workflow routing, bounded execution.

03
Verification & Governance

Independent review, CI gates, security controls, confidence, and failure handling.

OPEN SOURCEPMB + ACR + HEPublic evidence, not résumé adjectives.
OPEN-SOURCE AI GOVERNANCE◆15-AGENT LOCAL CODE REVIEW◆CONTEXT AUTHORITY + PROVENANCE◆LOCAL LLM INFERENCE◆HUMAN-IN-THE-LOOP AUTOMATION◆BUILD · GOVERN · VERIFY · SHIP◆

Selected AI + product work

The proof is in the systems.

I want the work to answer the hiring question. Can I reason about AI architecture, build the thing, govern it, integrate it, test it, and still think like a product owner? These projects are the evidence.

AI engineering · Context systems

Open source

Personal Memory Bank

Persistent project context with authority, freshness, provenance, diagnostics, hooks, review workflows, drift detection, and deterministic guardrails.

  • Context engineering
  • Governance
  • CLI
  • CI
Read case study

Local AI · Multi-agent tooling

Open source

AI Code Review Agent

A local multi-agent review system with specialist isolation, sanitized diffs, confidence-aware findings, security controls, and IDE/CLI integration.

  • Ollama
  • Agents
  • Review
  • Security
Read case study

AI systems · Research framework

Public research

Harness Engineering

My engineering framework for the layer around the model: context, memory, instructions, tools, orchestration, governance, evaluation, and execution boundaries.

  • AI architecture
  • Evaluation
  • Governance
  • Execution
Explore the framework

Product engineering · Live telemetry

Open source

PitLogic

A deployed analytics PWA with device integration, alarms, stall detection, finish-time prediction, offline behavior, and 169 passing tests.

  • React
  • Telemetry
  • PWA
  • Testing
Read case study

See Side Quest Atlas, Tipsy Bunghole, Bowling Tracker, and the full project set →

How I build AI

Models are powerful. Systems make them dependable.

Prompting matters. So do context authority, tool contracts, state, retries, approval gates, verification, privacy, degraded behavior, and whether anyone can tell what happened when the model is wrong.

That surrounding layer is where I spend a lot of my time.

Enter Harness Engineering →
BUILD

Get close to the code.

APIs, Python, orchestration, local models, data boundaries, developer workflows, and the actual implementation constraints.

GOVERN

Bound the machine.

Give automation explicit authority, deterministic controls, and human approval where consequences deserve judgment.

VERIFY

Do not let AI grade itself.

Use independent review, tests, CI, structured evidence, and observable failure instead of quiet confidence theater.

More than AI

I build because I like solving things.

The AI work is central professionally. The rest of the site proves the same operating style shows up everywhere: travel products, live event workflows, bowling analytics, BBQ telemetry, woodworking, and Scouting leadership.

LeadershipEagle Scout · SachemService, mentoring, preparedness, and decades in Scouting.MakerNolan's WorkshopJigs, tools, furniture, work surfaces, finishing, and sawdust.Product builderReal workflows firstSoftware when software fits. No agent required.