About

Who is writing all this, and what they have actually shipped.

Senior AI technical lead (CTO) of Booq (booqai.app), an AI content-production platform, and the engineer behind Kartez Contract (kartezcontract.com), an AI legal-advisory system. PhD candidate in Artificial Intelligence at Isfahan University of Technology, where he also earned his BSc in Computer Engineering and MSc in Artificial Intelligence. Five years of professional AI experience spanning large language models, chatbots, computer vision, deep learning, graph neural networks, question-answering systems, speech synthesis and recognition, data mining, analytics and business intelligence — backed by hands-on software engineering, DevOps, web and database work. Experienced in leading and delivering heavy, high-stakes AI projects end to end.

Where the work sits

Where frontier AI research (LLMs, RAG, GNNs) meets industrial-scale, production software architecture.

Not a researcher who also ships, and not an engineer who also reads papers — the useful position is the one in between, where a result has to survive a cost ceiling, a latency budget and a customer.

Range

The first professional role, in 2019, was a SIM-card hardware-management module driving AT commands in Qt and C++. The current one is technical leadership of an AI platform that dubs video into Persian. Everything between those two — MRI denoising, Persian speech synthesis for a national operator, three years of industrial AI taken from zero to a sold product, a retrieval-grounded legal advisor — is on the CV below.

Research

MSc thesis, Isfahan University of Technology, September 2025 — predicting gene expression from digital pathology images:

Gene-expression assays are slow, expensive and logistically painful. Whole-slide pathology images are fast and cheap. This thesis asks whether the second can stand in for the first — recovering molecular-biology signal directly from gigapixel pathology images, under a hard constraint on compute and latency. The approach combines graph neural networks, vision foundation models, large-language-model techniques adapted for this task, and statistical/probabilistic methods.

BSc final project, September 2022 — simulating human visual cognition and distinguishing real perception from hallucination:

A year-long study using generative models — autoencoders and GANs — to model how the human brain constructs visual understanding, and to distinguish genuine perception from hallucination.

That project was about perceptual hallucination in 2022, three years before model hallucination became the central problem of applied LLM work. The PhD, started in September 2025 at Isfahan University of Technology, is in progress; its research direction will be published here once it is settled, rather than described in advance.

A note on numbers

The case studies on this site carry no benchmark figures. That is deliberate. The measurements exist inside the projects, not in a public CV, and a number invented to fill a card would be the single most checkable false claim here. Each metric appears the moment it is measured and confirmed, and stays hidden until then.