Quentin McClellan

Founder · Solutions Architect · Platform Systems Builder

I build evidence-first systems that make messy operations inspectable.

My work combines operations discipline, software architecture, DevOps habits, workflow automation, and local-first infrastructure. The goal is practical: turn scattered artifacts and process steps into traceable evidence before making bigger automation or AI claims.

View StarkGrid case study What I build Engineering proof projects

What I build

Practical architecture for evidence, operations, and automation.

Governed ingestion

Repeatable intake patterns for files, scans, spreadsheets, and operational records with traceable identity and reviewable receipts.

Workflow orchestration

Guarded workflow lanes that separate operators, conductors, execution boundaries, policy checks, and read-only proof surfaces.

Source-backed observability

Dashboards, summaries, and portfolio-safe proof points that make it clear what happened and which evidence supports each claim.

Developer platform tooling

CLI-first scaffolding, schema-driven templates, validation commands, and Nexus artifact integration for repeatable project creation.

Evidence distillation

Structured pipelines that normalize operational logs, produce incident artifacts, and expose bounded evidence to AI clients safely.

Local-first infrastructure

Linux, Docker, scripting, repository discipline, and lab hardware arranged for controlled rehearsal, cost awareness, and explainability.

Featured project

StarkGrid

StarkGrid is a private MVP/demonstration platform for deterministic artifact ingestion, evidence review, and governed automation. Built in Java across seven active modules — Janus, Demeter, Argus, Themis, Hermes, Hephaestus, and Mnemosyne — running on Kafka and Docker Compose.

The system runs distributed across two lab nodes: Jarvis as the primary control-plane and Ultron as a GPU-capable remote worker. Athena (React) and athena-observer (FastAPI) provide read-only evidence review and DLQ visibility.

The case study demonstrates clear module ownership, deterministic processing boundaries, source-backed observability, and guardrailed claims — with explicit caveats for what has not been enabled.

Open the case study

Engineering Proof Projects

Four public projects demonstrating the underlying disciplines.

StarkGrid is the main platform. Forge, SRE Log Pipeline, Forge MCP, and Operator Evidence MCP are public projects that demonstrate repeatable scaffolding, evidence distillation, validation discipline, and safe AI-facing evidence interfaces.

Forge · v0.2.0

Developer Platform CLI

Python CLI for creating standardized project lanes with schema-driven templates, validation, Nexus artifact integration, and release workflows.

Proof points

  • Three project templates: python-worker, java-spring-service, node-dashboard
  • forge doctor validates local prerequisites
  • forge project inspect and Nexus template packaging
  • 28 tests; tagged release v0.2.0

Role fit: Developer tooling, platform engineering, release discipline

View Forge case study

SRE Log Pipeline · v0.2.0

Evidence Distiller

Python JSONL pipeline that turns noisy logs into structured incident evidence for debugging, incident review, and postmortem workflows.

Proof points

  • Zero runtime dependencies; timezone-aware benchmark-window filtering
  • Incident evidence schema v1.0; StarkGrid adapter
  • 8 test files covering batch, window, distillation, storage, and adapter behavior
  • Tagged release v0.2.0

Role fit: DevOps, SRE, observability, incident response

View SRE Log Pipeline case study

Forge MCP · v0.1.1

Developer Platform MCP Server

Read-only MCP server that exposes Forge repository knowledge — templates, validation commands, project structure — to AI clients without write access.

Proof points

  • 10-tool contract validated by test suite
  • No write or execution tools; request validation
  • Bounded responses; version v0.1.1

Role fit: AI tooling, developer platform, MCP integration

View Forge MCP

Operator Evidence MCP · v0.1.1

Incident Evidence MCP Server

Read-only MCP server that exposes bounded incident evidence and operational summaries to AI clients without write access or execution authority.

Proof points

  • 9-tool contract: evidence discovery, incident listing, summaries, trace samples
  • Slug/path validation; bounded results and truncation handling
  • Security boundary tested; version reporting via serverInfo
  • Version v0.1.1

Role fit: AI infrastructure, incident tooling, secure evidence retrieval

View Operator Evidence MCP

Contact

Connect with Quentin McClellan.

Use the professional domain email or LinkedIn profile for public portfolio follow-up.

Email: contact@quentinmcclellan.com

LinkedIn: linkedin.com/in/quentinmmcclellan