I will claude code setup and tutoring, mcp servers, ai agents and vibe coding help


About this gig
I will tell you honestly whether an hour will fix it.hether an hour will fix it.You are using Claude Code, Cursor or one of the AI builders, and it is doing maybe half of what you hoped. It forgets your conventions. It edits files you did not ask it to touch. It is confident exactly when it should not be.
Most of that is setup, not the model.
WHAT I DO
Set up your project so the AI stops guessing: written conventions it actually reads, the right files in front of it, and limits on what it is allowed to change.
Connect it to your own tools and data with MCP servers, so it stops inventing answers it could have looked up.
Split work across agents that plan, build and review, with checks before anything is called done.
Review what it already wrote and tell you plainly what is fine and what will bite you.
HOW IT RUNS
A live session over screen share, in English, Hindi or French, on your actual project. You keep the recording and a short written summary of what changed and why.
WHY ME
This is how I work every day, on software real people actually use, with two review gates before anything ships.
Message me with what you are using and what is going wrong.
Get to know Piyush K
I fix, build and deploy apps made with Lovable, Bolt, Replit and Base44
- FromIndia
- Member sinceDec 2021
Languages
Hindi, English, French
My Portfolio
Other Vibe Coding Services I Offer
FAQ
What if one hour is not enough?
Then we do not pretend it was. Tell me the problem before you order and I will say which package actually fits, or that you do not need me at all. If we run out of time mid-session I will finish the thread in writing afterwards rather than leave you half way.
I am not technical. Can you still help?
Yes. Describe the process in plain words: what happens today, and what you want to happen instead. I turn that into a plan with a fixed price and a delivery date before you order, so there are no surprises.
You have no reviews yet. Why should I risk it?
Fair question, and the risk sits with me rather than you. I am priced below what this work is worth precisely to earn a first review, I show you progress as I build rather than at the end, and if the first milestone is not what we agreed you do not pay for it.
Do I have to give you access to my code?
No. A screen share is enough, and you stay in control of the keyboard the whole time. If you would rather I made the changes myself, invite me to a branch and remove me afterwards. I never need your passwords or your API keys.
Why not just keep prompting until it works?
Sometimes you should. But there is a point where each fix breaks something else and the credits go faster than the progress. If you have been round that loop three times on the same bug, the loop is the problem, not the prompt. An hour usually costs less than the credits do.

