Context Engineering
Compression, pruning, dedup and view-layer optimization that keep agents sharp as conversations grow.
I work across Rust, TypeScript and Python — building context-engineering tools, agent infrastructure, GPU-accelerated audio DSP and full-stack frameworks. Dozens of projects shipped, one live SaaS, and a relentless bar for what ships.
followed by CEOs · PhD ML researchers · prominent builders across AI & tech
For 35+ years I've helped people turn ideas into real companies — across web, e-commerce and operations. We're at the beginning of the age of AI, and that experience is exactly the edge it takes to build for it well.
So I founded Engineered Automated Systems for Artificial Intelligence to ship the things I wished existed: Rust frameworks, agent and context infrastructure, and GPU audio tooling — all in the open. I care about craft, and about helping people leave the corporate grind to build something that's theirs.
Six domains, one through-line — production-grade systems engineering. Each is backed by shipped, open-source work, not slideware.
Compression, pruning, dedup and view-layer optimization that keep agents sharp as conversations grow.
MCP servers, lifecycle hooks and multi-agent orchestration — the protocols that wire tools to models.
GPU-accelerated, spectral-domain processing — loudness, restoration, mastering and deep analysis.
A Laravel-grade developer experience, re-engineered in Rust with batteries-included starter kits.
Tree-sitter parsing, symbol graphs and dependency analysis for assistant-grade codebase understanding.
Songwriting, prompt design and metadata systems for AI music generation.
A Rust framework and its two Tier-1 starter kits — plus the cognitive scaffolding that makes agents reason better.
A Laravel-inspired, full-stack framework for Rust. Routing, ORM, auth, mail and migrations — batteries included, borrow checker intact.
A production-shaped account story on day one — registration, email verification, password reset and profile management. Wired and tested.
The full product-site kit — marketing landing, docs pipeline, blog with RSS, a member community and role-based access control.
A toolkit for sharper decision-making through structured cognitive approaches — reusable reasoning scaffolds for more reliable agents.
A creator-owned music network where artists own their audience, their content and their payments — not a label, not a middleman. Designed and shipped end-to-end on Suprnova, with direct-to-fan payouts in 160+ cryptocurrencies.

MixLens runs on dedicated server infrastructure with GPU acceleration. It works in the spectral domain — full frequency-by-frequency processing, psychoacoustic modeling, and a mastering chain that decides on what it actually hears in your audio, not a preset table.





The tooling behind the work — pruning context, graphing code, speaking MCP, and keeping agents reliable. Open source, mostly Rust.
Safe Rust bindings for the Celemony ARA2 SDK — write ARA2 plugins that read DAW audio regions directly, in pure Rust.
A practical guide to the meta-tags that steer Suno AI — genre, mood, instruments, effects and song structure.
Silent-first context optimization for Pi — compresses, dedupes and prunes tool results automatically, with zero disruption to agent behavior.
Dynamic context pruning for the Hermes agent — keeps the window lean as conversations grow.
A Rust codebase-analysis library with Node & Python bindings — Tree-sitter parsing, symbol extraction and dependency graphs.
An MCP server for code intelligence — graph the codebase, feed the agent.
A standards-compliant Rust implementation of the Model Context Protocol — client, server, schema and transports.
A meta-cognitive workflow architecture for how agents plan, reason and self-correct.
A hook that fires on task-list completion to enforce best coding practices.
A high-performance, cross-platform process manager, inspired by PM2.
Thirty-five years building the systems and processes companies run on — now focused entirely on the AI era.