Software engineer · Gig Harbor, WA

I build AI that runs on your own machine.

Full-stack and systems work in TypeScript, React, Python and Rust: a GPU-rendered spatial workspace, on-device models in Web Workers, and the enablement work that gets a whole company using them. I also teach it. 200+ people trained, and a go-to-market story to go with the build.

EngineeringTypeScript, React, Electron, Rust and WebAssembly. Most recently a GPU-rendered spatial workspace with 5,000+ tests and performance gates behind it.
AI on the deviceSemantic search, OCR, speech-to-text and local LLMs running offline on the user's own machine, and on-premises for HIPAA and GDPR buyers.
Enablement & go-to-marketI teach what I build. Workshops, exec sessions, demos and positioning, for engineers and buyers both.

Project · local-first AI spatial workspace

A canvas where every document is still a real file.

Documents, spreadsheets, code, PDFs, calendars and 3D models live on one infinite canvas as fully editable items, in two-way sync with files on disk. A Rust/WebAssembly engine draws the canvas on the GPU, and the AI (semantic search, OCR, speech-to-text, a local LLM) runs on the machine, offline.

  • GPU renderer. wgpu/WebGPU with a WebGL2 fallback, drawing 1,000+ rich items at a 7.1 ms typical frame, with pan and zoom multitudes faster than the JavaScript DOM renderer it replaced.
  • On-device AI. Semantic search on ONNX Runtime Web, OCR, Whisper speech-to-text and a llama.cpp assistant, all in Web Workers so the canvas never drops a frame.
  • Files stay files. Dropped files stay linked to their items; items made in the app get real files on disk; dragging one out opens it in the right editor.
  • Built to stay honest. A pnpm monorepo with 5,000+ automated tests, CI and performance regression gates. Yjs CRDTs for conflict-free sync.
TypeScriptReactElectron RustWebAssemblyWebGPU Yjs CRDTsONNX RuntimeVitest
Theme
drag to pan · scroll to zoom
Drag the canvas or an itemScroll or pinch to zoom Click an item to open itSearch to filter

A browser-light recreation, built for this page in about 9 KB of JavaScript. It lays the cards out at whatever zoom you leave them at, so the type stays sharp, and drops to blocks when they are too small to read. The real workspace does this through a Rust/WebGPU engine with a WebGL2 fallback.

Sep 2026 – present · SchematicSense

Co-founder, Chief Design Officer & GTM

A pre-launch startup building on-premises AI chatbots with local LLM deployment, for healthcare, finance and legal organisations bound by HIPAA, GDPR, FERPA and COPPA. The buyers who cannot send a prompt to someone else's cloud.

Local LLM deploymentPyQt6HIPAA / GDPR Product designGo-to-market

The product

I designed the PyQt6 desktop application from scratch and own the product's UI and UX end to end.

The market

Developing go-to-market strategy and positioning for privacy-sensitive, regulated buyers, the same story I tell in a demo.

Mar 2026 – Sep 2026 · Denali Advanced Integration

AI Team Intern

Redmond, WA

Brought Claude into a company of engineers and operators, not as a pilot but as a working habit. Part platform work, part teaching, part showing executives what the numbers meant.

  • Model benchmarking. Evaluated Rainier-VL, Denali's open-weight 2.1B-parameter vision-language model for edge defect detection, which outperformed models up to 5× its size.
  • Internal tools. Built a full-stack note-taking app wired into an AI projects dashboard.
  • Bench automation and demos. Partnered with lead engineers on returns bench automation and presented technical demos to shareholders and clients.
Claude & Anthropic APIPrompt engineering LLM evaluationVision-language models Enablement

200+ per session

Led AI enablement workshops and large-format training with executive leadership in the room.

40%+ productivity

Directed tailored Claude projects for multiple teams and measured the lift rather than claiming it.

Months earlier

Built and maintained the Claude skills and plugins that accelerated the company's transition.

Feb 2025 – Mar 2026 · Auto Warehousing Company

Junior Programmer Analyst

Tacoma, WA

Operations managers had reports. What they wanted were answers. I built a Microsoft Copilot Studio agent that turned static business reports into a conversational analytics tool used by 40+ people across operations and IT. Hours of analysis down to minutes.

  • Computer vision. Wrote custom Python YOLO object detection for image and video recognition, cutting label scan time by 80% and reducing errors by over 15%.
  • Process automation. Partnered with technology and operations teams to design and propose automation initiatives across the yard.
Microsoft Copilot StudioPythonYOLO Computer visionPower Platform
01

Reports land

Nightly operational exports, the same files the team already trusted.

02

Agent grounds

Copilot Studio topics and knowledge sources mapped to the real report schema.

03

Anyone asks

Plain questions in Teams. No SQL, no pivot table, no waiting on IT.

04

Answer with source

Figures come back with the report and date they came from.

Operations assistant

An illustration of the agent's shape, with representative figures, not live company data.

May 2023 – Dec 2024 · University of Washington, Earth & Space Sciences

SeaFreeze

Application Development Intern · Seattle, WA

SeaFreeze computes the thermodynamic and elastic properties of water and ice polymorphs up to 100 GPa and 10,000 K, the physics behind the oceans under Europa and Enceladus. I redesigned the Python GUI around it for the dozen-plus researchers who use it daily, and delivered the software and presentations to NASA board members.

  • Faster 3D. Significantly improved responsiveness and rendering speed of the complex 3D property graphs researchers wait on.
  • Simpler workflows. Restructured a modular PyQt6 application around the questions scientists actually ask, for over a dozen researchers.
  • NASA audience. Delivered the software and presentations to NASA board members.
PythonPyQt6Matplotlib NumPy / SciPyPlanetary science

Schematic phase boundaries drawn for this page, in SeaFreeze's own palette and viridis colour map. The published model computes them from Gibbs local basis functions.

Ice IhIIIII V / VIliquid water 150250350 450550 Temperature (K) Pressure (MPa) 05001000 15002000

Hover or drag across the diagram to read a point.

Drag to rotate. Density of ice Ih over pressure and temperature.

Contact

Looking for engineering or technical sales roles.

B.S. Computer Science, University of Washington (Dec 2024). A.A. Computer Science, Tacoma Community College. Happy to walk through any of this live.

Or directly: bloomjack91@gmail.com 253-432-1152