Muhammad Jazab N.

Most of my work now is fixing .NET and Azure systems that were built fast with AI and then started breaking. Cursor, Claude Code and Lovable get a product 90% of the way. The last 10% is where it goes wrong: OAuth that fails when tokens expire, secrets sitting in code, tenants who can see each other's data, apps rejected by Apple or Google over data protection. That's the part I fix. 9 years in .NET and Azure • 4,700+ Upwork hours • 100% Job Success • Top Rated Plus 𝗔 𝗿𝗲𝗰𝗲𝗻𝘁 𝗲𝘅𝗮𝗺𝗽𝗹𝗲 A SaaS company wanted their Xero Practice Manager integration checked before launch. It looked finished. I found hardcoded credentials and an OAuth flow that broke as soon as tokens expired. We fixed both before a single customer touched it. 𝗪𝗵𝗮𝘁 𝗜 𝗴𝗲𝘁 𝗵𝗶𝗿𝗲𝗱 𝗳𝗼𝗿 • Reviewing AI-written code: auth holes, race conditions, missing tests and tenant data leaks, fixed before production • AI agents and MCP servers: Claude or ChatGPT working with your own APIs and data, with a person approving anything risky • App Store and Play Store rejections: unencrypted local data, missing permission flows, privacy policy gaps • Modernization: Framework 4.x to .NET 10, Xamarin to .NET MAUI • Slow SQL Server and PostgreSQL, Stripe, Xero, HubSpot and Microsoft Graph integrations, multi-tenant SaaS 𝗣𝗿𝗼𝗱𝘂𝗰𝘁𝘀 𝗜'𝘃𝗲 𝗯𝘂𝗶𝗹𝘁 • 𝗧𝗮𝗽 𝗟𝗼𝘆𝗮𝗹𝘁𝘆: digital stamp cards for cafés on .NET 10, with a working MCP server so an owner can ask Claude "who hasn't been in for a month" and get the answer from their own data • 𝗣𝗿𝗶𝘃𝗮𝘁𝗲 𝗔𝗜 𝗳𝗶𝘁𝗻𝗲𝘀𝘀 𝗰𝗼𝗮𝗰𝗵: a 2B model I fine-tuned on 15,532 rows, shrunk to 1.3 GB and running fully offline on the iPhone. It picks the right app tool 100% of the time on held-out tests, and no health data leaves the phone. 𝗛𝗼𝘄 𝗜 𝘄𝗼𝗿𝗸 I read the code before I quote, I tell you straight if something isn't worth doing, and I keep my project count low so I'm the one actually working on your system. Send me the error, the repo or the rejection email, and I'll tell you honestly what it'll take.