About

What I work on

Backend

Python services and APIs, built with FastAPI and Django. Clean architecture, tests, and handling the cases where something downstream fails rather than assuming it will not.

Applied AI

LLM applications: RAG, prompt and response handling, and working with more than one provider. The interesting part is rarely the model call — it is everything around it that decides whether the answer is right and whether the system survives a provider being slow or unavailable.

AI-assisted development

Using coding agents and LLMs as part of how the work gets done: the prompting, the context they need to be useful, and reading what comes back closely enough to know whether it is right before it goes anywhere near a codebase.

Evaluation

Working out whether a model is actually correct, including the cases where it is confidently wrong. This is what my contract work with Mercor is: adversarial evaluation, verified golden answers, and rubrics that make a failure reproducible instead of arguable.

Turning an idea into a product

Taking something from a prototype that demonstrates an idea to software that other people can actually use. Most of that work is problem-solving: understanding the problem, choosing an approach that will hold, and dealing with what breaks along the way.

Where I am

I hold a B.Sc. in Computer Science from the University of Guilan. Since finishing my degree my time has gone into coursework, independent learning, personal projects, and contract work evaluating frontier LLMs.

I'd rather be judged on what I've built than on a count of years, which is why the work is written up with the reasoning behind it rather than summarised as a list of technologies.

I'm based in Rasht, Iran, and open to backend and AI engineering roles — particularly remote and international ones. I read and write English and Persian.

Elsewhere

CV Contact