Audit
I read the repo and tell you in plain English what is real, fragile, fake, and worth fixing.
Services
For the moment your AI-built app gets 80% of the way there, then turns into a locked room with no windows.
You are not waking up at 3am because of stack traces. You are staring at a chatbot that almost fixes the bug, breaks two other things, and cannot explain the code it just created. I turn that almost-product into software you can ship.
What this is good for
- Turning 80% vibecoded demos into shippable product work.
- Explaining what the code is doing in normal language before changing it.
- Replacing chatbot guesswork with actual engineering judgment.
Engagement shapes
Project builds
Short audits
Workshop sessions
Fractional advisory
Fix
I take the broken last mile and turn it into working software with less lying in the code.
Teach
I show you how to use AI tools without letting the tools drive the product.
01
Software Engineering
Full-stack implementation for the app AI helped you start but cannot responsibly finish. Features, services, integrations, data flow, and the production details the bot keeps hand-waving.
You have the idea and the demo. Now it has to become software.
- Last-mile implementation
- Production fixes
- Maintainable code
02
Frontend Engineering
React, TypeScript, forms, auth screens, state cleanup, SSR, and UI behavior for interfaces that looked fine until you clicked anything twice.
Your UI is one prompt away from breaking itself.
- Stable UI
- State cleanup
- SSR-friendly pages
03
Consulting
A blunt read on what you actually have, what is fake, what is dangerous, and what needs to happen next before you spend another week arguing with an autocomplete box.
You are tired of not knowing whether the app is close or cooked.
- Plain-English diagnosis
- Scope cuts
- Build plan
04
VibeCode Unfucking
Triage for the 80% app stuck in the endless almost-fixed loop: duplicated state, haunted workflows, mystery files, broken assumptions, and code nobody on the call can explain.
The demo worked. The last mile did not.
- Damage report
- Priority fixes
- Cleaner code
05
Effective AI Use Workshops
Practical sessions for founders, operators, and teams using AI to build without blindly accepting code they cannot read, test, or recover from.
You want leverage without making the chatbot your CTO.
- AI workflow
- Prompt discipline
- Review habits
06
AI Systems and Agent Tooling
LLM workflows, RAG, MCP, model routing, agent runtimes, and tool-use systems for when the AI feature itself needs to behave like software, not a haunted text box.
You are past the toy phase and the weird failures are getting expensive.
- Agent flows
- Tool contracts
- Evaluation loops
07
Architecture Review
A direct pass over the folder structure, data flow, dependencies, failure modes, and places where the AI invented structure that only looks like engineering.
You do not know what good code looks like, but you can feel this is not it.
- System notes
- Failure modes
- Refactor sequence
08
Performance and SSR Repair
Slow pages, broken hydration, weird routes, cache mistakes, SEO-critical rendering, and the classic problem where localhost keeps lying to you.
The app works on your machine because your machine is being polite.
- Render fixes
- Load improvements
- Deployment checks
09
Technical Leadership
Staff-level execution help for founders and teams who need someone to translate the mess, make decisions, review code, and keep the build from dissolving into prompt archaeology.
You need engineering judgment before the app becomes folklore.
- Review standards
- Delivery rhythm
- Sharper ownership
Not sure which one fits?
Send the half-working thing. If it is 80% real and 20% impossible to finish, that is exactly the part I can help with.