Hire
Senior applied AI tooling — embedded engagements, methodology audits, internal agent infrastructure setup, fractional architect retainer.
Get in touchWhat we build
Opinionated, LLM-native machinery for engineering workflows. Context engine, agent orchestrator, methodology — production-tested over 9 months on a 136K-line Rust codebase, agents shipping merged PRs daily. Not a framework, not a SaaS. Machinery designed for the LLM to operate within and improve. Configurability is overhead the LLM has to pay on every task; opinions remove discovery cost.
Engagements
Prototype build — Unity
Vibe-coded prototype or MVP in Unity / C# for game ideas, interactive experiences, or internal tools. Output is a clean Unity project your team takes over and extends after the build. Twelve years Unity contracting plus our agent stack means a prototype lands faster than a junior-led build, in code your engineers can read.
Contact for ratePrototype build — JS / web
Vibe-coded web app, game, or viz prototype in your JS / TS stack. Hand-off-ready: your engineers pick it up and extend in their stack. Same agent toolkit, same velocity advantage — different runtime.
Quick picks by use case
- Web app →
- 3D game →
- 2D game →
- Data / scientific viz →
- WebGPU-first today →
Embedded engineer
Senior agentic engineer embedded with your team, shipping features against your codebase using our toolkit. Duration scoped with you — weeks to months.
Contact for rateMethodology audit
Review of your Claude Code / agent setup, with recommendations report and prioritized fix list. Typically 1-2 weeks.
Contact for rateAI tooling architecture
Bootstrap your team's internal AI tooling — CLAUDE.md, context docs, skills, lint rules, role definitions, hands-on training. Typically 4-6 weeks.
Contact for rateFractional AI architect
On retainer for AI tooling decisions — design reviews, architecture calls, hiring guidance. Monthly engagement, ongoing.
Contact for rateWhat you get
- Shipped commits in your language and conventions.
- A methodology layer — CLAUDE.md, context docs, skills, lint rules — that lives in your repo and your team owns post-engagement.
- ~3-5x velocity vs vanilla Claude Code during the engagement, after ~1.5-2 weeks of domain-onboarding setup.
- Agents that improve the methodology with every task — your context docs grow, lint rules sharpen, role definitions adapt.
How we work
- Setup First 1-2 weeks: we onboard your domain into the context engine — ingest from your codebase, Confluence, internal wikis, Slack channels you share. Configure agent roles for your stack. Set up CLAUDE.md with your conventions.
- Engagement Our agents ship features through the same workflow as a human PR — planning, isolated worktrees, two-tier review gates. Every change goes through review before merge.
- Handoff Methodology layer lives in your repo. Your team + Claude Code can keep using and extending it after the engagement ends. No black-box dependency.
The fit matrix
- Best fit
- Embedded engineer engagements, Claude Code shops, dev-tools / AI-native gaming / mid-size product teams.
- Partial fit
- Copilot shops (methodology + context engine port; runner adapter is one-time work). Audit-shape engagements work across any agent tool.
- Wrong fit
- Maintaining LangChain pipelines, hourly-rate code dictation, procurement-locked enterprises requiring approved frameworks. We refer those out — different lane.