Zack Meacham

Software engineer · Applied AI

I build AI-powered products, and use agentic tooling to build them well.

Applied AI is the throughline: LLM-powered features people actually use, and the agentic workflows (MCP servers, custom skills, RAG) that change how the engineering gets done. The discipline underneath comes from mission-critical systems work at Boeing.

View selected work Resume

Proof

  • LLM-powered products

    Ledger / Vera: a civic AI assistant where every claim cites its source; provenance is enforced in the type system, not bolted on. Cofounder & CTO.

  • Agentic AI workflows

    Custom skills, MCP servers, RAG pipelines, and tool-using agents, built and run day-to-day, including as a primary internal resource for frontier-AI adoption at Boeing.

  • Mission-critical systems

    Test infrastructure at Boeing across Ada, C++, LabVIEW, and hardware. The rigor that makes the AI work hold up under real constraints.

  • Full-stack delivery

    Web, desktop, and CLI products taken from zero to real usage.

Selected work

Three flagship case studies.

All work

  1. Flagship · applied AI product

    2026 · Pre-Alpha · Cofounder & CTO

    Ledger / Vera

    A civic AI assistant where every claim cites its official source: Vera answers only from a closed set of grounded tools, and an un-sourced fact cannot be constructed in the type system. A federal data foundation built, with state-level coverage activating behind a fail-closed bar.

    • Next.js
    • TypeScript
    • Supabase
    • Anthropic API
    • Vitest
    • Playwright
    Read case study →

  2. Flagship · independent systems

    2026 · Active build · Solo

    Milodex

    A solo-built autonomous trading system where every order clears a fail-closed risk chokepoint no code path can bypass: exactly-once execution drain, append-only event store, a 17-check veto layer, and CLI plus Qt desktop operator surfaces. Paper-only, enforced in code; governed by 58 ADRs and a custom Opus risk-reviewer.

    • Python
    • Alpaca
    • SQLite
    • Parquet
    • PySide6 / Qt
    • pytest
    Read case study → GitHub

  3. Flagship · professional engineering

    2022 – present · Shipped & ongoing · public-safe

    Test Infrastructure at Boeing

    Mission-critical test infrastructure for a strategic defense guidance system, written public-safe. Three concurrent efforts spanning Ada, LabVIEW, hardware integration, and stakeholder-level systems work (one originated, one led, one built from scratch), plus internal frontier-AI adoption across two organizations.

    • Ada
    • C++
    • LabVIEW
    • Hardware integration
    • Requirements engineering
    Read case study →

How I work

How I work.

  • Systems thinking

    Depth before noise: understand the system, then make the smallest high-leverage move.

  • Agentic workflows

    I build with tool-using agents and frontier models day-to-day, not as a novelty but as how the engineering actually gets done.

  • Earned autonomy

    Autonomy is earned, not assumed. Milodex refuses orders by default and won't move live capital without an explicit, human-approved promotion; in Ledger an un-sourced fact cannot be constructed.

  • Restrained product judgment

    Architecture with a purpose: clear interfaces, small blast radius, honest failure modes, built for maintainability and operator trust.

Experience

Full role history and skills.

Open resume

Software Engineer 2 at Boeing since 2022 and cofounder and CTO at Ledger, with personal systems and tooling shipped alongside.

Contact

Reach out.

Contact

Email is the quickest way to reach me. Happy to talk about applied AI, agentic engineering, product and systems work, or something interesting you think I should be building.

Get in touch Email