Safi M.

I'm a .NET architect with 13+ years building microservices and multi-tenant SaaS, and lately I've been building AI on top of them. I've shipped systems that handle over $18M in transactions and RAG assistants that run in production, not demos. I'm also CTO at Centangle Interactive, where I run a 50 person engineering team across 14+ projects. I only take a few contracts on Upwork, and I do the architecture work myself. You won't get handed to a junior. Some of what I've built: A multi-tenant SaaS platform for refinery operations. I scaled it out of a monolith into 8 microservices with 200+ endpoints and 150+ database tables, using RabbitMQ and MassTransit for events, Ocelot as the gateway, and OpenIddict for identity with a custom role resolution system. It tracks budgets on projects worth millions. A financial ERP running across 90+ branches and 20+ organisations, processing more than $18M in deposits and withdrawals. It's offline first, syncs across branches, and consolidates reporting for the whole network. An AI assistant that's live on Google Play. RAG with vector search, HyDE query expansion, ingestion for PDFs, images and audio, streaming chat over SSE, and routing across multiple LLM providers. A natural language to SQL engine that turns plain English questions into tenant-scoped queries. Every generated query gets security filters injected automatically, so a user can never read another tenant's data. It has semantic caching and supports more than one LLM provider. A mentorship platform for the British Council, where I was lead architect. Microservices, Zoom integration, real time chat, GDPR compliance. I'm probably a good fit if you have a .NET monolith you want broken into microservices without a rewrite from scratch, a multi-tenant SaaS with permissions or data isolation problems, or a system where correctness matters because money moves through it. I'm also a good fit if you want AI built properly into a real product: a RAG assistant over your own documents, a chatbot that answers questions from your database in plain English, or an LLM pipeline that has to handle multi-tenant security and real users rather than a demo. And if your problem is the data itself, slow queries, a schema that's fighting you, or reporting that has to stay correct across dozens of locations, that's the same work to me. Tools I work in: ASP.NET Core 8, C#, microservices, RabbitMQ, MassTransit, Ocelot, OpenIddict, IdentityServer, EF Core, SQL Server, Docker, Clean Architecture, CQRS, DDD, React, React Native, Python, FastAPI, OpenAI, Claude, Gemini, RAG, vector search, n8n. Tell me what's breaking or what you're trying to build, and I'll tell you honestly whether I'm the right person for it.