I work across frontend, backend, cloud, AI, and infrastructure to build systems end to end.
I'm Faraz Ali Ahmad, a Software Engineer. I have experience building web applications, cloud infrastructure, and AI systems.
I've worked on web applications, internal tools, AI systems, and client products. At Morvion, I worked across products including Dreilokale, Incontro, the company website, CRM, and their AI systems, contributing to new features, performance improvements, integrations, and production systems.
A timeline of my work experience and the impact I've created along the way.
Software Engineer (Freelancer)
Software Engineer
Founding Engineer
Machine Learning Trainer Intern
A blend of technical expertise and problem solving to build scalable, real world solutions.
Where I build interfaces, interactions, and everything users see and use.
Where I build the APIs, services, and backend logic behind the application.
Working with Relational and non-relational databases, and data models to build data-driven applications
Building AI applications with LLMs, RAG, agents, and modern AI tooling.
Setting up the infrastructure, deployments, CI/CD, and everything behind the scenes.
The languages I use to build, experiment, and solve problems.
fkasdkfhkjasdhf
asdfjkkahsdfjkh sfhakjsfd asjfh jahfj akfhlasdfas df
sdjfhkadshflksdfhkhsd
Website performance is one of those things that looks simple on the surface and turns out to be a mix of a dozen small decisions underneath. In one of my recent projects, I improved the Lighthouse performance score from 59 to 99. I also reduced the initial data downloaded on page load from around 103 MB to 2.8 MB. That change alone made the site feel dramatically faster, and it reminded me of an important truth: performance is usually won by removing unnecessary work, not by adding clever tricks.
A deep dive into state machines, tool-calling pipelines, structured outputs, and fallback patterns when deploying LLM agents to production systems.
Concrete techniques to slash JavaScript bundles by 97%, eliminate Cumulative Layout Shift (CLS), and achieve sub-50ms TTFB at the edge.
Website performance is one of those things that looks simple on the surface and turns out to be a mix of a dozen small decisions underneath. In one of my recent projects, I improved the Lighthouse performance score from 59 to 99. I also reduced the initial data downloaded on page load from around 103 MB to 2.8 MB. That change alone made the site feel dramatically faster, and it reminded me of an important truth: performance is usually won by removing unnecessary work, not by adding clever tricks.
A deep dive into state machines, tool-calling pipelines, structured outputs, and fallback patterns when deploying LLM agents to production systems.
Concrete techniques to slash JavaScript bundles by 97%, eliminate Cumulative Layout Shift (CLS), and achieve sub-50ms TTFB at the edge.
I'm always open to interesting projects, freelance work, or just a good tech conversation.