Mahdi MolaviI am Mahdi Molavi, a senior backend engineer with more than ten years of production work in C#, .NET and SQL Server, and a researcher in tensor methods for large-scale data. I take on selected projects in the four areas below.
Websites and web applications
Personal, company and product websites that load fast, work well on phones and are ready for search engines from the first day. I build them on .NET with a modern front end, or on an existing CMS when that suits the client better. Bilingual sites with a proper right-to-left Persian version are part of that; this site, with its English pages and its separate Persian section, is an example.
For reference, this site's home page scores 100 for accessibility, best practices and SEO in Lighthouse, with a largest contentful paint of about 2.2 seconds on a throttled mobile profile (measured in October 2026).
Backends, APIs and performance
ASP.NET Core services, SQL Server tuning, Redis caching and message queues for systems that have to stay fast under load. I designed a Redis Sentinel setup that served 45,000 to 70,000 concurrent exam users, and I architected an adaptive testing platform used by about 89,000 students. Typical work includes finding the real bottleneck, fixing slow queries and indexes, adding caching where it pays off, and load testing before a peak.
I have also written up how I work in four practical guides: SQL Server performance tuning, load testing with Locust, caching in ASP.NET Core and reliable messaging with RabbitMQ, all in the backend guides.
Search and AI visibility
The technical groundwork that lets search engines and AI assistants find a site and understand who is behind it: a clean heading structure, structured data (JSON-LD) for people, organisations, services and articles, sitemaps, hreflang for multilingual sites, an llms.txt file, Core Web Vitals, and setup of Google Search Console, Bing Webmaster Tools and IndexNow. Nobody can promise a ranking; what I make sure of is that the site is fast, readable and consistent enough to be trusted.
Tensor methods and machine learning
Consulting and implementation for problems where the data has more than two dimensions, such as images, video, sensor streams and brain imaging: tensor decompositions like t-SVD and tc-SVD, low-rank compression, and tensor-based classification and clustering. My M.Sc. thesis at Tarbiat Modares University was on this topic. The Tensor Notes series explains it step by step, and a MATLAB implementation of tc-SVD is on GitHub.
Starting a project
Email mhdmlv93@gmail.com or message me on LinkedIn with a few lines about what you need, the deadline, and what a good result looks like for you. I work remotely and can correspond in English, Persian, Turkish and Azerbaijani.