Performance Metrics
Solo developer. AI-augmented workflow. Measured output.
Real data from git history across the xsubi ecosystem — no estimates, no embellishment.
Headline Numbers
Output vs. Industry Baseline
Measured against published developer productivity benchmarks
AI Model Orchestration
Each model routed to its strength — strategy separated from execution
- Architecture decisions
- Use case authoring
- Tradeoff analysis
- Code review
- Task routing
- Quality gates
- Queue management
- PR automation
- Code generation
- File scaffolding
- Test writing
- Implementation
- Feature branches
- Pull requests
- Passing tests
- Merged to develop
Throughput
Daily Commit Velocity
Last 30 days — commits across all repositories
Project Breakdown
| Project | Domain | Commits | Source Lines | Tests | Stack | Progress |
|---|---|---|---|---|---|---|
| xsubi-docs | Architecture / UCs | 102 | 17,808 | — | Markdown | 60%
Active |
| qforge | Dev Tools / MCP | 37 | 18,326 | 463 | PythonMarkdownJavaScript | 65%
Active |
| xsubi-infra | IaC / DevOps | 30 | 4,395 | — | YAMLBashMarkdown | 50%
Active |
| robonet | Distributed Systems | 27 | 147,794 | 2,990 | PythonMarkdownJavaScriptBash | 85%
Alpha |
| home-lab-operations | Lab Ops | 10 | 4,362 | — | C#MarkdownYAMLBash | 0%
Planned |
| xsubi-learning | Learning Platform | 8 | 3,581 | — | C#JavaScriptMarkdown | 30%
Early |
| xsubi-resume | AI / Automation | 8 | 17,391 | — | PythonMarkdownBash | 50%
Active |
| xsubi-host | Infrastructure / API | 6 | 327 | — | Markdown | 40%
Early |
| xsubi-website | Web / Platform | 5 | 64,078 | — | C#TypeScriptMarkdownJavaScriptPython | 55%
Active |
| xsubi-games | Game Servers | 4 | 8,412 | 186 | TypeScriptMarkdown | 35%
Early |
| odoo-campground-bensons-resort | ERP / Demo | 3 | 1,264 | — | PythonMarkdownJavaScript | 0%
Planned |
| xmark | Version Registry | 2 | 2,204 | 27 | PythonBashMarkdown | 80%
Alpha |
| xsubi-bench | Benchmarking | 1 | 104 | — | Markdown | 20%
Planned |
| xsubi-drones | Drone Hardware | 1 | 291 | — | Markdown | 15%
Planned |
| xsubi-farm | Server Farm | 1 | 76 | — | Markdown | 15%
Planned |
| xsubi-support | Support / Ops | 1 | 52 | — | Markdown | 20%
Planned |
| xsubi-turrets | Turret Systems | 1 | 212 | — | Markdown | 15%
Planned |
| xsubi-vtt | Virtual Tabletop | 1 | 74 | — | Markdown | 20%
Planned |
Technical Coverage
Domain Expertise
Distributed Systems
Peer-to-peer mesh networking, gossip protocols, failure detection, leader election, cross-subnet federation. roboNet: autonomous drone coordination over UDP.
Cloud Infrastructure
Self-hosted multi-hypervisor platform. KVM + Hyper-V provisioning, Kubernetes orchestration, Cloudflare networking. Zero cloud vendor dependency.
AI-Augmented Development
Custom AI tooling (xsubi-forge, qforge) for autonomous task dispatch. MCP server integration. Multi-model workflow architecture (Claude, GPT, Grok).
Observed Patterns
Emergent engineering patterns from building the xsubi ecosystem
Test Density Tracks Blast Radius
Projects that other systems depend on have the highest test-to-code ratios. xsubi-forge: 1 test per 9 lines. Leaf projects (www, games): zero tests needed. Test investment follows risk, not ritual.
Burst-Consolidate Rhythm
Productive days cluster: 30-35 commit peaks followed by 1-4 commit valleys. High-output days build new capability; low-output days stabilize, test, and refactor. Trying to sustain peak throughput daily degrades quality.
Rewrite > Refactor at Threshold
robonet-a reached 66 commits before architectural limits made further iteration expensive. robonet-b (clean rewrite) hit equivalent functionality in 8 commits with better architecture. Rewriting was 8x more efficient than continuing to patch.
Zero-Dep Core, Optional Edges
Both xsubi-forge and qforge have zero required external dependencies. Optional features (MCP server, web UI) use optional-dependencies with guarded imports. Core reliability never depends on third-party stability.
Filesystem as IPC
Multiple tools communicate through the filesystem (inbox JSON, status files, history logs) rather than network protocols. Debuggable with cat. Survives process crashes. No serialization framework needed.
Documentation Scales with Complexity
xsubi-docs has the most commits (84) of any repo — more than the distributed system itself. Complex systems require more documentation effort than code effort. Architecture docs, use cases, and roadmaps prevent re-solving solved problems.
AI Separation of Concerns
Opus (architecture, planning, evaluation) never writes implementation code. Sonnet (code generation, scaffolding) never makes architectural decisions. Same separation-of-concerns principle applied to AI tooling that applies to software layers.
Shared Tokens, Separate Apps
xsubi-website and xsubi-portal are independent Blazor apps that look like one product through a shared CSS design token system. Visual unity without code coupling. Either app can be deployed, scaled, or modified independently.
- AI doesn't replace engineering judgment — it amplifies execution speed
- Multi-model routing outperforms single-model usage by 3–5x
- Autonomous pipelines (forge) eliminate context-switching overhead
- Output quality is maintained through automated testing (1,451+ tests)
- The multiplier compounds: faster iteration → faster learning → better architecture
Methodology
Every metric on this page is sourced directly from git log across the xsubi ecosystem. No self-reported estimates.
Claude Code (Opus for architecture, Sonnet for implementation), GPT, Grok — used as tools, not crutches. Every line is reviewed, tested, and owned.
1,451 automated tests across Python projects. xsubi-forge alone has 704 tests for 6,381 lines — a 1:9 test-to-code ratio.
One person across every layer: distributed systems, hypervisor APIs, web UIs, CI/CD pipelines, infrastructure automation, documentation, and AI tooling.
What's Next
Build trajectory through 2026
Site Development
xsubi.com build progress — click each phase for details