Connect your departments
- The problem
- Departments chase each other by phone and the information lives in scattered Excel files.
- The fix
- One integrated system where every order's status is clear to every department at any moment.

Designed with creativity, delivered with expertise
Digital Transformation · Enterprise Systems · AI Automation
s.ashouri.s@gmail.com · +98 910 484 4127 · linkedin.com/in/saeid-ashouri-0a28a4193 · Iran
I help manufacturers and service companies bring scattered departmental work into one integrated system, see accurate reports on time, and automate repetitive work with AI.
Optical system online
Four kinds of work, each aimed at a problem I've seen in real companies.
The first step is a short conversation. Describe the problem, and we'll see which part is worth building first.
Book an intro callA promise is a promiseDeadlines included
Metsa is a long-established aluminum window and door manufacturer. Departments coordinated by phone, and much of the staff's time went into chasing each other's work. Production had no clear view of upcoming orders, which made planning the line hard. Design work also relied on several separately licensed programs.
After interviewing about 12 people across every department, we chose not to buy an off-the-shelf ERP and designed a system around how this company actually works. Now every order’s status is clear to every department, the old design software has been rebuilt inside the system, the site crew records measurements on a phone, and each department has its own dashboard.
The aim is fewer calls and less rework, more accurate production planning, and a company ready to sell in other provinces. Next, the same system can be offered to smaller workshops so they can produce as part of Metsa’s production network.
Contractor and product lead. I design and build every part of the system, using AI tools to move faster. Mohammadreza Ashrafi works with me on the finance assistant.
New AI tools appear every day, and finding the right one, comparing prices and keeping up with the news takes real time. Cheshm.ai set out to bring all of it together in one complete reference.
A directory that introduces and categorizes AI tools, with advanced search to find them. Around it we built a prompt marketplace, AI calculators, model sales, a marketplace for ready-made automations from around the world, and a news section that updates itself.
A business owner can see in one place which AI tools exist for their work, what they cost and how to use them.
Co-founder and technical product lead. I designed the system's technical structure, its services and the automated data pipelines, and took the product to launch.
An online shop for 3D-printed products, started on almost no capital by repairing broken printers. It sold across several social and messaging apps, and monthly customers grew from 170 to about 340 in six months.
Before software, I spent years on drones, electronics and vehicle systems. That road is told on the backstory page.
Open the backstory →For technical readers, an overview of the system's parts and how they connect. The dashed module is still being built.
A crypto trading signal that made money by hand was tested before automation, with dozens of trading bots, two exchanges and a full year of data.
A precise market-data recorder and a live arena where dozens of trading bots competed with real fees and risk limits, without real money. AI could only veto a trade.
Above all, the signal predicts volatility, and flagged coins were 2.7–3× more likely to spike sharply. One rule held in all 12 walk-forward months, at +0.83% per trade. In part of the market execution costs outweigh that edge, and a simple filter on the bid-ask spread lifts net returns by about 20%.
Two research papers, each replicated on two independent exchanges, a package in which every number is rebuilt by a script, and a clear, documented investment decision.
Research lead. I owned the decisions and checked every result, and the programming was done with the AI tool Claude Code.
A rule that showed +120.6% over 8 days came to about zero over a full year.
Hardware taught me to measureSoftware didn’t change that
Metsa Aluminum · Qom, Iran
Self-employed · Remote
Agenx · algorithmic trading research
Cheshm.ai · Tehran
Setare Penhan Pars Co. · Tehran
Pixel 3D · E-Commerce
Civil Aviation Technology College
Before proposing anything, I talk with managers and staff across departments to learn how the company really works. A system built on that understanding gets used instead of shelved.
I build with the help of AI tools. Design, technical decisions and quality control stay with me, and AI speeds up the building. For the client that means a usable version sooner.
Before software I was an avionics engineer and spent years on drones and vehicle systems, so production floors and workshops are familiar ground.
Conversations with managers and staff in every department, recording how work actually happens, before any design.
Design and quality control stay with me, and AI tools speed up the building.
The features of old specialist software rebuilt inside your own system, without separate licenses.
Scattered Excel files and old systems moved into one database, with users checking and approving.
Assistants that work on your own data, like Metsa's financial report assistant, now in testing.
An engineering background in drones, electronics and vehicles, and at home on production floors.
For technical readers. Most of the work is in TypeScript and Python, with AI tools part of every step.
The AI tools and models I use every day.

Master's Degree, Entrepreneurship / New Venture
2024 — 2026GPA 19 / 20
B.Sc. Electrical Engineering, Aviation Electronics
2019 — 2023GPA 16 / 20Advanced Course Certificate. Proficient in reading, upper-intermediate in writing, speaking and listening.
I once built from scrapNow I build for factories
If you want your departments working in one system, reports you can trust on time, or repetitive work handed to AI, get in touch for an intro call.