Skip to content
View mknoth197's full-sized avatar

Block or report mknoth197

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
mknoth197/README.md

Hey there! πŸ‘‹ I'm Mitch Knoth

5x AWS Certified | ☁️ Cloud Software Engineer at John Deere | AI-Native Delivery Advocate

Harness engineering for the AI-era SDLC

πŸ™‹β€β™‚οΈ About Me

mitchell:
  role: Cloud Software Engineer @ John Deere
  current_focus: Harness engineering for the AI-era SDLC
  working_on:
    - Agentic workflows & agent context engineering
    - Telemetry-backed metrics for AI engineering productivity
    - Evals-driven harness + model selection at enterprise scale
  passions:
    - AI Agents & Agentic Workflows
    - Developer Experience & Platform Engineering
    - Confident Continuous Delivery
  background: 5x AWS Certified β€” cloud infrastructure & networking
  mission: Help engineering teams ship safer, smarter, and faster β€” with agents as first-class teammates
  • πŸ€– Exploring AI-native software delivery β€” agentic workflows, agent context engineering, and telemetry-backed metrics for AI productivity
  • 🧰 Focused on harness engineering β€” the thesis that the models are good enough; the harness (instructions, hooks, context, evals) is what's lagging
  • πŸš€ Pushing confident continuous delivery β€” codifying engineering policy as instructions + deterministic hooks so agent-written code ships safely from PR to production
  • ☁️ 5x AWS Certified β€” a cloud infrastructure & networking background underneath the AI-native work

πŸ”­ What I'm Working On

  • 🧠 Investigating AI-native delivery patterns and measuring real-world AI engineering productivity (DXI + telemetry-backed signals)
  • 🧰 Defining the harness β€” normalizing instructions, context, and hooks across coding agents (Copilot, Claude Code, Codex, Gemini CLI) so every tool starts from equal footing
  • πŸ“Š Running evals to choose the right harness + model objectively β€” large-scale, policy-aware experiments (Promptfoo) instead of "this one just feels better"
  • πŸ” Agent session handoffs β€” a cross-session memory lifecycle engine + GitHub-Issues epic layer for zero-memory-loss context recovery
  • 🧱 Building a compounding knowledge base β€” an AI-maintained "LLM wiki" with on-device semantic search (QMD) and knowledge-graph analysis (Graphify)

πŸ› οΈ Tech Stack

Languages & Markup

Python TypeScript JavaScript Bash YAML Markdown

AI & Agentic Engineering

GitHub Copilot Claude Code OpenAI Codex Gemini CLI MCP Promptfoo Graphify QMD Obsidian

DevOps & Continuous Delivery

GitHub Actions Docker Dev Containers Renovate Biome Backstage Git Trunk-Based Dev

Cloud & Infrastructure (background β€” 5x AWS Certified)

AWS Terraform CloudFormation

πŸ’¬ "The best CI/CD pipeline is the one nobody has to think about."

Thanks for stopping by! πŸ˜„

Pinned Loading

  1. harness-haircut harness-haircut Public

    My personal collection of differentiating harness configurations

    TypeScript

  2. evidence-gated-delivery evidence-gated-delivery Public

    Evidence-backed Research, Plan, Implement, and Review workflow for agentic engineering.

    Python