Systems engineer, technical artist, and design engineer. One workshop, many routes.
I build across AI and local-model infrastructure, agent tooling and evaluation, compilers, native and browser rendering, seeded neural graphics, poster design, color, and AI-assisted design workflows. Project Telos maps fourteen flagship engines under one Flywheel thesis. Their maturity varies, but the shared habit is simple: map the system, make the surface, test the claim, and leave a usable artifact.
The banner names the ambition, not a blanket maturity claim. Some tools can run anywhere with few or no dependencies; others are Windows-native, GPU-facing, pre-1.0, or research-only. Some are released packages, some are public beta, and some are active R&D. Each project carries its own status below.
Start here: Project Telos · Portfolio · Resume · CV · LinkedIn
Explore: research · papers · Studio · the person behind it
Seattle, WA · Rust · Python · C++23 · ORCID 0009-0001-7175-5393 · open to paid employment, contract and project work, technical partnerships, research and academic learning paths, and practical IT.
| Lane | Start with | Current shape |
|---|---|---|
| AI and local-model infrastructure | flywheel, relay | Harness, endpoint, routing, failover, and evaluation work. Active R&D plus a 0.1.0 source prototype; benchmark conclusions stay scoped to the task set that produced them. |
| Agent tooling | telos, index, gather, forum, crucible | A mixed-maturity toolchain for context, research intake, orchestration, evaluation, and human/model workspaces. |
| Compilers and developer systems | BuildLang | Rust-built typed-effects compiler. The C execution path and HLSL/GLSL output are the current core; other backends and linear types remain explicitly experimental. |
| Design, graphics, and generated media | Studio Engine, Build Color, Elder ENB | Browser-native poster composition with measured critique, seeded neural graphics, native rendering, color workflows, and AI-assisted iteration beside an established public graphics project. |
| Color and calibration | Build Color | A 1.0.2 beta color-science workbench for spaces, HDR tone mapping, appearance models, difference metrics, ICC, and LUT workflows. |
| Research, learning, and release tooling | Learn, papers, release toolkit, emet | Accountable learning workflows, public papers, source and provenance systems, package and CI checks, release surfaces, and a byte-level integrity witness. |
The badges pull versions, CI state, and downloads from registries and GitHub on page load. Static maturity wording is labeled and bounded by project rather than promoted into one ecosystem claim.
| Tool | What it does | Maturity | The receipt that matters |
|---|---|---|---|
| flywheel | Runs local and hosted model routes through the same harness and oracle-backed evaluation paths. | Research · active R&D. No portfolio-wide model-superiority claim; each result belongs to its recorded task set, endpoint, and budget. | Cross-harness manifests, endpoint receipts, raw benchmark artifacts, and explicit non-conclusions. |
| telos | Shared human/model workbench, MCP surfaces, creative tools, and four proof lanes through one CLI. | Public work · 0.2.0 pre-1.0 source-registry package. Tested surface; interfaces may move and npm publishing is operator-gated. | Re-checkable agent, research, visual, and build packets with explicit non-claims. |
| index | Maps a repo or workspace into an atlas, context envelope, symbol graph, or verified wiki. | Public work · 2.9.0 beta. Zero runtime dependencies. | File-and-line evidence, typed omission receipts, and freshness checks. |
| gather | Captures web, video, papers, PDFs, browser/OCR/audio, and structured sources into research packets. | Public work · 1.6.1 release. | Provenance and digest verification stay attached to each captured item. |
| forum | Routes multi-agent work through replayable ledgers, context budgets, gates, and campaigns. | Public work · 1.13.0 versioned public package. | Hash-chained bodies and records of who did what, under which route and constraint. |
| crucible | Registers a thesis, steelmans it, measures it, and emits a bounded verdict. | Public work · 1.2.0 versioned public package. | The verdict is recomputed from the recorded measurement rather than accepted from prose. |
| emet | Re-derives byte-level integrity facts without making trust or release decisions. | Public work · 1.1.0 package and release; frozen 1.0.0 core spec. Four same-author implementations share the core; receipt and rebind coverage is capability-specific. | A different-author implementation is still the open bar for demonstrated independent re-derivability. |
| buildlang | Compiles typed-effects source through C and emits HLSL/GLSL shader source. | Public work · 1.2.0 source manifest. C, effects, HLSL/GLSL, and receipts are core; SPIR-V, LLVM, WASM, Rust, native-ISA backends, GPU dispatch, and linear types are experimental. | Backend maturity is stated per target instead of hidden behind one compiler-wide label. |
| learn | Turns source material into coursework, retrieval practice, spaced repetition, and graded records. | Public work · 1.6.0 source version. Zero runtime dependencies. | mastery() is derived from the learner's recorded practice. |
Together, the two tables name all fourteen flagships.
| Project | Current role | Maturity |
|---|---|---|
| relay | Coding-agent and endpoint ladder for local servers, authenticated CLIs, and configured APIs. | Public work · 0.1.0 source prototype. Git install; scopes writes and execution behind explicit flags. |
| plexus | Discovers what agent tools emit and consume, exports the capability graph, and proposes inspectable pipeline routes. | Plexus 0.2.0 public source. Git install; unmatched inputs and cycles remain visible. |
| mneme | Local agent memory with provenance, reproducible ranking, and drift checks. | Public work · 0.1.0 source prototype. Git install; PyPI release is not claimed yet. |
| studio-engine | Generates replayable shader, audio, motion, and raster artifacts from a seed. | Research · 0.2.0 pre-1.0 engine. APIs may move. |
| build-color | Color spaces, HDR tone mapping, appearance models, difference metrics, ICC, and LUT workflows. | Public work · 1.0.2 beta. A workbench and toolkit, not a physical measurement instrument. |
flowchart LR
problem["real problem"] --> models["AI / local-model workflows"]
problem --> systems["compilers / developer systems"]
problem --> visuals["poster / neural / rendering / color"]
problem --> research["research / learning / release"]
models --> tools["flywheel · relay · forum · telos"]
systems --> compiler["BuildLang · index · plexus"]
visuals --> art["Studio Engine · Build Color"]
research --> evidence["gather · crucible · emet · papers"]
tools --> artifact["usable artifact"]
compiler --> artifact
art --> artifact
evidence --> artifact
artifact --> human["human review and next loop"]
The projects connect where their interfaces are real. They do not need to share one verdict vocabulary, deployment target, license, or maturity stage to belong to the same workshop. Rigor is the floor; it is not the only room.
These are not profile decorations; they are small doors into the workbench.
# map a workspace into one HTML file
pip install index-graph
index atlas --root /path/to/workspace --format html --out atlas.html
# replay an agent route, then watch the ledger catch a tampered result
git clone https://github.com/HarperZ9/forum && cd forum
python examples/demo.py
# force a verdict: does a confident claim survive the measurement?
git clone https://github.com/HarperZ9/crucible && cd crucible
python examples/demo.py
# one frozen core spec plus capability-specific conformance lanes
git clone https://github.com/HarperZ9/emet && cd emet
python conformance/run.py membrane.pyMore surfaces: Studio (visual), catalog (atlas), flagships overview, research lanes.
The loop.
Scan the real project. Read the old sessions. Find the current state. Pull the docs, repos, tests, demos, and market evidence into the same room. Build the smallest tool that changes the situation. Use it immediately. Let it fail in public or near-public conditions. Fold the failure back into the product. Commit, push, verify, repeat.
That loop is the personality. I get impatient when work becomes posture, when tools are protected from real use, or when a claim cannot be made to stand next to its source. I am more interested in the moment where the thing breaks and becomes better than the moment where it first sounds impressive.
Why accountability.
I do not write about accountability because I think I am naturally accountable. I write about it because I know how easy it is to dodge the mirror: blame the room, overclaim the work, take shortcuts, want credit before earning it, or confuse intensity with progress.
That is why several Telos tools put a claim beside its source, let a model say
UNVERIFIABLE, or make an action leave a receipt. Those mechanisms belong
where they are useful; they are not a universal costume for the compiler,
graphics, color, or product work. The personal version of this is on
person.html.
The pressure I put on tools.
- Dogfood it: if the tool is for developers, run it on real repositories.
- Adversarially test it: make the smallest failure case and keep it.
- Make it public when it can be: ship the repo, demo, issue, receipt, or page.
- Keep the art alive: rigor can still have color, rhythm, naming, and motion.
- Do not sand off the ambition: narrow the next step without pretending the larger project stopped mattering.
I came up without a CS degree or industry certification. The credential is the public trail: shipped crates, a published VS Code extension, Elder ENB (a Skyrim graphics project whose public career materials report more than 900,000 downloads), open repositories, and tools that can be cloned, run, and argued with.
The work that shaped me.
- Technical Networking Support, Xbox Division: TCP/IP, DNS, NAT, router configuration. The first hard lesson that a correct answer is not useful until another person can act on it.
- Operations Manager / Lead Arborist, family business: field work, client relations, scheduling, proposals, budgets, safety procedures. Accountability that is not abstract.
- Freelance technical writing and consulting: API guides, security and compliance documentation, onboarding material. Explaining systems without exposing client internals.
- Independent engineering since 2023: local-model infrastructure, agent tools, compilers, graphics, color science, research tooling, release systems, and public demos under Project Telos.
The projects that changed the shape of the work.
- Elder ENB: two years of public releases and named editions; current public career materials report more than 900,000 downloads. Taught taste, iteration, users, and the difference between a pretty frame and a maintained system.
- Native graphics lineage: D3D11/HLSL renderers, proxy-DLL interception, mid-frame compute dispatch, ACES/AgX tone mapping, TAA, SSR, SSGI, GTAO, volumetrics, ImGui tools, CMake/vcpkg, shared-memory IPC.
- Build Color: a color-science workbench. Color spaces, HDR tone mapping, perceptual difference metrics, chromatic adaptation, ICC profiles, gamut work, color-vision simulation, 3D LUTs.
- BuildLang: a typed-effects language and compiler line. Lexing, parsing, checking, effects, lifetimes, C FFI, C lowering, editor support, explicit maturity labels for unfinished parts.
- Local-model and agent infrastructure: endpoint adapters, harness comparisons, verifier-guided search, context maps, agent ledgers, and bounded evaluation artifacts.
- Design and graphics workflows: browser-native poster composition, measured contrast, overlap, and placement critique, seed-authored neural graphics, deterministic rendering, color workflows, and AI-assisted iteration and export.
- Project Telos: the workshop that lets these lines stay distinct while sharing maps, tests, interfaces, and release discipline.
I am open to full-time, part-time, contract, and project work across systems engineering, AI and evaluation infrastructure, developer tooling, compilers, graphics and rendering, design engineering, technical art, implementation, research operations, QA, technical writing, and practical IT. The strongest fit is work that needs someone to enter an ambiguous system, map it quickly, build the missing surface, and leave behind something another person can run and inspect.
- Local-model endpoints, agent harnesses, evaluations, context systems, and developer tools.
- Compilers, language tooling, native systems, and difficult integration work.
- Real-time graphics, shaders, procedural media, color science, poster systems, and AI-assisted design workflows.
- Research infrastructure, benchmark design, packaging, CI, release evidence, documentation, and technical support.
I am also interested in research, academic, fellowship, residency, mentorship, and continued learning opportunities where I can contribute real engineering while learning from people working at the frontier.
I also work with organizations as an independent technical peer. Depending on the tool and its public license, an engagement can cover a paid pilot, subcontracted delivery, implementation and support, managed access, a private deployment, or term licensing for a defined scope. For MIT-licensed tools, the fee covers deployment, integration, customization, managed operation, training, or support, not code-use permission already granted by the license.
A good first engagement starts with a concrete repository, workflow, benchmark, or design problem and ends with explicit acceptance criteria, working artifacts, and a usable handoff.
Portfolio · Resume · CV · LinkedIn
If you would rather evaluate the work than read positioning, run one of these public seams against a real workflow and report where it fails.
- Test gather intake
- Test index maps
- Test forum ledgers
- Test crucible checks
- Test the telos surface
- emet's highest-leverage ask: a different-author implementation from SPEC.md alone, in any language, passing
conformance/vectors.json. That converts re-derivability from asserted to demonstrated.
This README is part of the workbench. It has a local verifier and CI, and stays deliberately static: no badge wall beyond what each tool actually earns, no visitor counter, no dashboard that silently rots. The banner is generated art: a seeded flow field in the Project Telos spectrum, the same family the site draws live in your browser.
git status --short
python scripts/check_profile_surface.pyClone it, run it, try to break it.