For an app that was built with AI — vibe-coded, generated or AI-assisted — and is now broken, stuck at 80%, or not safe to launch, the right hire is usually one experienced engineer who reads the code before changing it — not an agency that quotes a rebuild, and not a marketplace where the cheapest bid wins. This page is the honest version of that decision: when each option is right, what to look for, what it costs, and what to send.
The right developer depends on what you actually need. Find yours — each links to the page written for it.
Every prompt fixes one thing and breaks another. You need someone to find the cause once, not prompt harder. That is debugging work — hours or days, not weeks.
The demo works; payments, email, logins and the awkward cases don’t. You need someone to take an honest inventory and close the last mile without rebuilding the first 80%.
Usually a difference between the two environments — settings, database, build — not a code problem. Small, specific, and the most common reason people get in touch.
Before real users and real data arrive: who can see whose records, what happens when things fail, what a bad actor could do. A review with a prioritised list, not a rewrite.
A contractor, a co-founder, an agency or an AI session from six months ago. You need someone to take it over properly: map it before anyone touches it, then a plan.
The code is fragile enough that every change feels like a risk. The fix is usually structural and smaller than it looks.
Say so in the first message. Production outages go to the front of the queue, and I’ll want logs and hosting access, not just the code.
Each is right for something. Here is what each is actually good at, and where it goes wrong for AI-generated apps specifically.
“We were stuck close to launch. Umaiz helped identify what actually needed fixing instead of suggesting a complete rewrite.”
Most developers have never worked on generated code, and some will quote a rewrite before they have read it. Five things that separate the ones who will actually fix it.
A price before anyone has opened the repository is a guess. The honest answer to ‘how much?’ is ‘let me look first’ — followed by a fixed number.
‘We’ll rebuild it properly’ is the most expensive sentence in software. You want someone who reproduces the failure and traces it before proposing anything.
Cursor, Lovable, Bolt, Replit, v0 and Base44 each fail in their own recognisable ways. Someone who has seen those patterns finds the problem in hours rather than days.
If the fix comes back as a wall of code with no explanation, you depend on them for the next one too. You should understand what happened and why.
Your repository, your hosting, your accounts, your data. Nothing should live only on their machine or behind their login.
Any one of these is a reason to keep looking.
Send the link, or just describe what’s wrong. I’ll tell you what I’d fix first and what it should cost — before you commit to anything.
This is how it runs with me. A good solo or freelance engineer should look similar.
A URL, a repository, a screenshot, a Loom, or just a description on WhatsApp. It doesn’t need to be tidy and you don’t need to know what’s wrong.
I read the code, reproduce the failure and come back with what I found, what I’d fix first, what can stay — and a fixed price for that work. You decide before anything starts.
No account manager, no hand-offs. The person you message is the person in the code. Remote, worldwide, with enough timezone overlap that questions get answered the same day.
Fixes land in your repository as commits you can read. If something goes wrong you know exactly which change did it — which is never true of a rewrite.
A one-off fix, a finish-and-launch, or ongoing help — your call. Nothing in the setup depends on me being available next month.
Umaiz — software engineer & architect, 7+ years across mobile, backend, web and AI-powered systems: Flutter, Node.js, FastAPI, Python, APIs, databases and cloud. Remote, worldwide, working directly with founders on AI-generated and AI-assisted applications.
The longer version is on the About page. The short version: I read before I write, I’d rather tell you something inconvenient early than discover it together later, and I leave things in a state where you don’t need me.
“Very easy to work with, technically strong, and straight to the point. He solved what we couldn’t figure out.”
There is no price list, because there is no such thing as an average AI-generated app. What there is: a fixed quote after a first look, so you know the cost before any work begins, and nothing changes partway through. Small, well-defined fixes are small quotes. A finish-and-launch is scoped from an inventory of what is actually missing.
The cheapest option is almost never the rebuild, and almost never the lowest marketplace bid — it is the person who finds the three real problems in an app that felt like it had thirty.
Send what you have. If you don’t have all of this, send the rest and I’ll tell you what else I need.
Try the AI once or twice more — sometimes that is enough. If two rounds haven’t fixed it, the model is missing context it can’t get from your description, and each further attempt adds an untested change. That is the point where a developer who reads the code, reproduces the failure and fixes the cause is cheaper than another week of prompting.
For fixing, finishing or reviewing an app that already exists, one experienced engineer is usually the better hire: they read the code before quoting, fix the cause rather than rebuilding, and you talk to them directly. An agency makes sense when you need a whole team from day one. A marketplace bid makes sense for a tiny, well-defined task — generated code with an unknown fault is not that.
It depends on the work, so I don’t publish a price list. After a first look I quote a fixed price for the diagnosis, and any fix is scoped from the findings — so you know the cost before any work begins, and nothing changes partway through.
Yes — all of them, plus Claude, Copilot and ChatGPT output. I don’t much care which tool generated the code. What matters is the application underneath it: the architecture, APIs, database, dependencies, and where things are breaking. Lovable and Bolt projects tend to need deployment and backend work; Cursor and Copilot projects more often need the seams between generated pieces checked.
Yes — I work remotely by default and have done for years. Timezone overlap is usually a scheduling question rather than a blocker.
Less than you’d think. A repository or a deployed URL, some idea of what’s broken, and access to whatever is failing — logs, the hosting dashboard, an error message. If you don’t have all of that, send what you do have and I’ll tell you what else I need.
The service
The app half-works. You've described the bug to the AI six times, and each fix moves the problem somewhere else. At some point the loop stops being cheaper than reading the code.
A practical method for debugging AI-generated code when prompting has stopped helping: reproduce it, read the seams, check the assumptions, and fix the cause.
Read
8 min read
The last 20% of an AI-built app is integrations, edge cases and deployment. Take an honest inventory, wire real payments and auth, and get it live.
Read
9 min read
It might need an engineer.
If you’ve spent hours asking AI to fix the same problem and you’re going in circles, stop fighting the code. Send me what you’ve got — I’ll figure out what’s wrong and tell you what I’d do next.
No cleanup. No judgment. The messy version is usually the useful version.