Hire me
AI-generated code debugging and code review
Your AI coding tool wrote most of the app, and now something is wrong: a bug nobody can trace, slow pages, or code you are not sure is safe. I review it, find the cause and fix it.
Quick answer
I debug and review code written with AI tools such as Cursor, Claude Code, Lovable, Bolt.new and GitHub Copilot: tracing bugs to their cause, fixing security issues, speeding up slow queries and APIs, and leaving tests so the fix stays fixed. I fixed random JWT logouts on a 2M+ user platform and scaled a backend from 200K to 1M+ users with about 60% faster APIs.
What I build
Bug hunting
Reproduce the problem, trace it to the cause and fix it, instead of patching symptoms the AI tool keeps reintroducing.
Code review
A written review of structure, duplication, error handling and risky patterns, ranked by what matters most.
Security review
Secrets, authentication, permissions, injection and dependency issues found and fixed.
Performance
Slow queries, missing indexes, N+1 requests and heavy pages measured and fixed.
Tests
Tests around the fixed code and the main user flows, so future AI edits do not break them again.
Refactoring
Tangled AI-generated code made clear enough for you and your AI tools to keep working on.
Results on real products
TheCryptoBasic
2M+ users
Fixed random logouts caused by JWT session bugs such as refresh races between tabs; the full write-up is on this site.
CV Maker – Resume Builder
200K → 1M+ users
Found and fixed the slow paths in a live backend: about 60% faster APIs with MongoDB indexing and Redis caching.
DOP
300M+ users
Fixed timeouts and RPC downtime in a live product used at very large scale.
Tools I use
CursorClaude CodeGitHub CopilotLovableBolt.newNode.jsTypeScriptReactNext.jsPythonMongoDBPostgresRedisJestPlaywright
How a project works
- Scope call. We talk through the goal, the current state of the product and what success looks like.
- Written plan. You get fixed milestones, delivery dates and a clear price before any work starts.
- Build in small pieces. I ship in small steps with regular progress updates, so you always see what is built.
- Tested delivery. Delivery includes tests, documentation, handover and free support afterwards if something breaks.
Frequently asked questions
Can you fix bugs in code written by Cursor or Claude Code?
Yes. AI-written code is still code: I read it, reproduce the bug, find the cause and fix it in your repository.
Why does my AI tool keep reintroducing the same bug?
Usually because the cause is elsewhere, in shared state, a race or a wrong assumption. Fixing the cause and adding a test stops the loop.
What does an AI code review include?
A written report of bugs, security issues, performance problems and structure, ranked by risk, plus fixes for the most important items.
Is AI-generated code safe to put in production?
It can be, after review. Common risks are exposed keys, missing permission checks, unsafe database rules and missing input validation.
Do you work in our repository?
Yes. I work in your Git repository with small, reviewed commits or pull requests.
Which languages and frameworks?
Mainly TypeScript and JavaScript (React, Next.js, Node.js), Python, Swift, and MongoDB, Postgres or Supabase databases.
Can you speed up a slow app?
Yes. I measure first, then fix the real bottlenecks: queries, indexes, caching, bundle size and API design.
Do you write tests?
Yes, around the fixed code and your main user flows, so later changes do not silently break them.
How quickly can you start?
Usually within a day or two. Send the repository and a description of the problem for a first look.
How is it priced?
Small fixes are hourly; a full review or larger fix is a fixed price after a first look at the code.