DEV Community: Azamat Narzulloyev

Building Real AI Systems From Uzbekistan — The Engineering Journey Behind Innosoft.uz

Azamat Narzulloyev — Fri, 06 Feb 2026 15:49:12 +0000

In the global tech industry, most stories come from Silicon Valley, Europe, or large Asian tech hubs. But interesting engineering challenges exist everywhere — including Central Asia.

Over the last few years, while building Innosoft.uz, we discovered that creating real AI-powered systems is less about hype and more about solving practical constraints: performance, scalability, and integration with messy real-world workflows.

This article is not a marketing pitch. Instead, I want to share how we approach engineering problems, why we made certain architectural decisions, and what we learned building AI-focused products from Uzbekistan.

Starting With Problems, Not Technology

Many teams begin with tools:

“We want to use AI.”
“We want computer vision.”
“We want automation.”

We started differently.

Most local businesses didn’t need another website or dashboard. They needed systems that could:

reduce manual work
make decisions faster
operate reliably under real conditions

The goal was not to use AI for its own sake, but to build systems that improve operational efficiency.

This changed how we designed everything from architecture to deployment.

Engineering Philosophy: Performance First

One of the biggest mistakes we noticed in early-stage development is prioritizing features over performance.

When building AI systems, performance is not a later optimization — it is the core requirement.

Some principles we adopted early:

Design for asynchronous workflows from day one.
Avoid heavy abstractions unless necessary.
Measure latency continuously.
Optimize data pipelines before adding new features.

For example, in real-time video processing pipelines, small inefficiencies multiply quickly. A few milliseconds lost per frame becomes unusable latency at scale.

Building Real-Time AI Systems

One of the core engineering challenges we faced involved multi-stream computer vision systems.

Typical demo implementations work fine on single inputs. Reality looks different:

multiple camera streams
unstable network conditions
limited hardware resources
unpredictable workloads

To solve this, we moved toward:

dynamic batching strategies
GPU acceleration using ONNX and TensorRT
asynchronous processing pipelines
intelligent load balancing between CPU and GPU tasks

The lesson was simple:

AI models are only part of the system. The surrounding infrastructure determines whether they actually work in production.

AI Agents Beyond Chatbots

The industry currently treats AI agents mostly as conversational tools.

Our experience showed that businesses benefit more from action-oriented AI.

Instead of just answering questions, we focused on building agents that:

integrate directly with business processes
trigger workflows automatically
analyze incoming data streams
assist decision-making rather than replace it

One example involved Telegram-based AI assistants designed for operational support rather than general conversation.

This shift from “chat” to “action” significantly improved real-world usefulness.

Lessons From Building Outside Major Tech Hubs

Operating from Uzbekistan introduced unique challenges:

Limited local infrastructure compared to larger markets
Fewer specialized AI engineering communities
Need to self-optimize learning and experimentation

But it also provided advantages:

strong focus on efficiency
engineering-driven problem solving
less distraction from trends and hype cycles

In many ways, constraints forced better technical decisions.

Architecture Decisions That Worked Well

Some patterns proved consistently valuable:

API-first system design
modular microservices without excessive fragmentation
careful monitoring from early stages
investing time into deployment automation

Interestingly, the biggest performance improvements rarely came from complex algorithms — they came from simplifying data flow.

Why Branding Matters for Engineering Teams

Many engineering-focused companies ignore branding, assuming that good code speaks for itself.

In reality, visibility matters.

Publishing technical content helps:

attract collaborators
build credibility
document engineering thinking
create long-term trust signals

Platforms like dev.to allow smaller teams to share real experiences without needing large marketing budgets.

Looking Forward

The next phase for Innosoft.uz is not just building products but building systems that scale globally while remaining efficient.

Key areas of interest:

real-time AI infrastructure
scalable AI agents
performance-focused software architecture

Over the last few years, while building Innosoft.uz
, we discovered that creating real AI-powered systems

Professional Web Sayt Yaratish - To'liq Qo'llanma

Azamat Narzulloyev — Wed, 24 Sep 2025 17:47:31 +0000

Web sayt yaratish - bu faqat kod yozish emas, balki strategik yondashuv talab qiladigan murakkab jarayon. Muvaffaqiyatli loyiha uchun quyidagi bosqichlarni ketma-ket bajarishingiz kerak.

1. Loyihani Rejalashtirish va Tahlil Qilish

Maqsad va Auditoriyani Aniqlash

Sayt maqsadi nima? - Biznes, portfolio, blog, e-commerce
Target auditoriya kimlar? - Yoshi, qiziqishlari, texnik darajasi
Asosiy xususiyatlar - Foydalanuvchilar nimani qilishi kerak?

Raqobatchilarni O'rganish

Sohangizda boshqa saytlarni tahlil qiling
Ularning kuchli va zaif tomonlarini aniqlang
O'zingizning unique value proposition yarating

2. Dizayn va UX/UI Rejalashtirish

Wireframe va Mockup Yaratish

Tavsiya qilinadigan vositalar:
- Figma (bepul va professional)
- Adobe XD
- Sketch (Mac uchun)
- Canva (oddiy dizaynlar uchun)

Responsive Dizayn Prinsiplari

Mobile-first yondashish - Kichik ekranlardan boshlang
Flexible grid system - 12-column grid ishlatish
Breakpoints - 320px, 768px, 1024px, 1200px

Ranglar va Tipografiya

2-3 ta asosiy rang tanlang
Contrast ratio kamida 4.5:1 bo'lsin
Web-safe shriftlarni ishlatish (Google Fonts)

3. Texnik Yechim Tanlash

Frontend Texnologiyalar

Boshlang'ich darajasi uchun:

HTML5 + CSS3 + Vanilla JavaScript
- Sodda va tushunarli
- Hech qanday framework kerak emas
- Tez yuklash

O'rta daraja uchun:

React.js yoki Vue.js
- Component-based architecture
- Virtual DOM
- Katta community support

Professional daraja:

Next.js (React-based) yoki Nuxt.js (Vue-based)
- Server-side rendering
- SEO optimizatsiya
- Performance optimization

Backend Tanlash (Agar kerak bo'lsa)

Oddiy saytlar uchun:

Static Site Generators: Gatsby, Hugo, Jekyll
Headless CMS: Strapi, Contentful

Murakkab loyihalar uchun:

Node.js + Express.js
- JavaScript ecosystem
- NPM packages
- Real-time features

Django yoki Flask (Python)
- Rapid development
- Admin panel
- Security features

4. Development Jarayoni

Loyiha Strukturasini Yaratish

my-website/
├── src/
│   ├── components/
│   ├── pages/
│   ├── assets/
│   │   ├── images/
│   │   ├── css/
│   │   └── js/
├── public/
├── package.json
└── README.md

Version Control (Git)

# Loyihani boshlash
git init
git add .
git commit -m "Initial commit"

# GitHub/GitLab ga yuklash
git remote add origin [repository-url]
git push -u origin main

Kodlash Best Practices

HTML Best Practices:

<!DOCTYPE html>
<html lang="uz">
<head>
    <meta charset="UTF-8">
    <meta name="viewport" content="width=device-width, initial-scale=1.0">
    <title>Aniq va qisqa title</title>
    <meta name="description" content="150 belgigacha tavsif">
</head>

CSS Best Practices:

/* BEM methodology ishlatish */
.header {}
.header__navigation {}
.header__navigation--active {}

/* CSS Variables */
:root {
    --primary-color: #007bff;
    --font-family: 'Inter', sans-serif;
}

5. Performance Optimization

Rasm Optimizatsiyasi

WebP formatini ishlatish
Lazy loading qo'llash
Responsive images (srcset)

CSS va JavaScript Optimization

// Minification va compression
// Gzip yoki Brotli compression
// Critical CSS inline qilish
// JavaScript chunking

Caching Strategiyasi

Browser Cache Headers:
- Static files: 1 yil
- HTML files: 5 daqiqa
- API responses: 15 daqiqa

6. SEO va Accessibility

Technical SEO

<!-- Open Graph tags -->
<meta property="og:title" content="Sahifa nomi">
<meta property="og:description" content="Sahifa tavsifi">
<meta property="og:image" content="rasm-url.jpg">

<!-- Schema markup -->
<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "WebSite",
  "name": "Sayt nomi"
}
</script>

Accessibility (A11y)

Semantic HTML ishlatish
Alt attributes rasmlar uchun
ARIA labels va roles
Keyboard navigation support

7. Testing va Quality Assurance

Turli Xil Testlar

// Unit testing (Jest)
test('function works correctly', () => {
    expect(myFunction()).toBe(expectedResult);
});

// Integration testing
// E2E testing (Cypress, Playwright)

Cross-browser Testing

Chrome, Firefox, Safari, Edge
Mobile browsers (iOS Safari, Chrome Mobile)
Accessibility testing tools

8. Hosting va Deployment

Static Saytlar Uchun (Bepul):

Netlify - Git integration, automatic deployment
Vercel - Next.js uchun eng yaxshi
GitHub Pages - GitHub repository'dan to'g'ridan-to'g'ri
Firebase Hosting - Google'dan

Dynamic Saytlar Uchun:

DigitalOcean - $5/oydan
AWS EC2 - Scalable solution
Heroku - Beginner-friendly
Railway - Modern alternative

Domain va SSL

# Custom domain ulash
# Let's Encrypt SSL certificate (bepul)
# CDN setup (CloudFlare)

9. Post-Launch va Monitoring

Analytics O'rnatish

// Google Analytics 4
// Yandex.Metrica (O'zbekiston uchun)
// Performance monitoring (PageSpeed Insights)

Security Measures

HTTPS enforcing
Content Security Policy (CSP)
Regular updates va patches
Backup strategiyasi

Performance Monitoring

Tools:
- Google PageSpeed Insights
- GTmetrix  
- Lighthouse
- WebPageTest

10. Davomiy Rivojlantirish

User Feedback

Contact forms
User surveys
Analytics data analysis
A/B testing

Regular Maintenance

Weekly:
- Content updates
- Security updates
- Performance check

Monthly:
- Analytics review
- SEO audit
- User feedback analysis

Quarterly:
- Major updates
- Feature additions
- Technology updates

Xulosa

Web sayt yaratish - bu bir martalik ish emas, balki davomiy jarayon. Muvaffaqiyat uchun:

Rejalashtirish - Vaqt sarflab, to'g'ri strategiya ishlab chiqing
Foydalanuvchi tajribasi - Har doim user-centric yondashish
Performance - Tez va samarali sayt yaratish
SEO - Qidiruv tizimlarida ko'rinish
Continuous improvement - Doimiy takomillashtirish

Esda tuting: Perfect sayt yo'q, faqat doimiy yaxshilanayotgan sayt bor!

Ushbu qo'llanma bilan professional va zamonaviy web sayt yaratishingiz mumkin. Har bir bosqichni puxta bajarib, o'z vaqtida foydalanuvchilaringizga yordam bering.
manba innosoft

Sayt yaratish bosqichlari: G‘oya Dizayn Kod Ishga tushirish

Azamat Narzulloyev — Sat, 06 Sep 2025 13:46:14 +0000

Professional Veb-Sayt Yaratish Bosqichlari: To‘liq Yo‘riqnoma

Bugungi raqamli dunyoda biznes yoki shaxsiy brend uchun professional veb-saytga ega bo‘lish muvaffaqiyatning muhim omilidir. Ammo ko‘pchilik “Sayt yaratish jarayoni qanday kechadi?” degan savolga duch keladi. Ushbu maqolada Innosoft jamoasi tajribasiga asoslanib, veb-sayt yaratishning asosiy bosqichlarini aniq, tushunarli va mukammal tarzda tushuntiramiz. Bu yo‘riqnoma orqali siz web development jarayonini A dan Z gacha o‘rganasiz.

Rejalashtirish va Tahlil

Har bir muvaffaqiyatli loyiha aniq reja va chuqur tahlildan boshlanadi. Bu bosqich loyihaning poydevorini shakllantiradi.

Maqsadlarni aniqlash: Sayt qanday maqsadda ishlatiladi? Masalan, onlayn savdo do‘koni, shaxsiy brending, blog yoki xizmatlar taqdimoti.

Maqsadli auditoriyani aniqlash: Saytga kimlar tashrif buyuradi? Ular qanday ehtiyojlarga ega? Yosh, jins, qiziqishlar va boshqa omillar hisobga olinadi.

Raqobatchilarni tahlil qilish: Raqiblarning saytlari qanday dizayn, funksionallik va marketing strategiyalariga ega? Ularning kuchli va zaif tomonlari o‘rganiladi.

Texnik talablar: Saytning funksional imkoniyatlari, masalan, to‘lov tizimlari, foydalanuvchi hisoblari yoki boshqa maxsus xususiyatlar aniqlanadi.

📌 Natija: Bu bosqichda Sayt xaritasi (Site Map) tuziladi, ya’ni saytning asosiy sahifalari va ularning o‘zaro bog‘lanishi. Shuningdek, kontent rejasi tayyorlanadi, unda qanday matnlar, rasmlar va videolar kerakligi ko‘rsatiladi.

Dizayn (UI/UX)

Dizayn bosqichi saytning foydalanuvchilar uchun qulay va jozibali bo‘lishini ta’minlaydi. Bu jarayon ikki asosiy qismdan iborat: UI (User Interface) va UX (User Experience).

Wireframe: Saytning oddiy chizmasi yoki sxemasi. Bu sahifalarning tuzilishi va elementlarning joylashuvini aniqlaydi.

UI dizayn: Ranglar palitrasi, shriftlar, tugmalar, ikonalar va boshqa vizual elementlar tanlanadi. Brendning uslubi (logo, ranglar, shior) hisobga olinadi.

UX dizayn: Foydalanuvchi oqimi (user flow) ishlab chiqiladi. Bu navigatsiyaning soddaligi, qidiruv tizimining qulayligi va foydalanuvchi tajribasini osonlashtirishni o‘z ichiga oladi.

Mobil moslashuvchanlik: Sayt har qanday qurilmada (mobil telefon, planshet, kompyuter) mukammal ko‘rinishi va ishlashi kerak (responsive yoki adaptive dizayn).

🔑 Qoidalar:

Minimalist dizayn: keraksiz elementlardan qochish.

Tez yuklanish: optimallashtirilgan tasvirlar va engil kod.

Brendga moslik: sayt brend imidjini aks ettirishi kerak.

Frontend Ishlab Chiqish

Frontend — bu foydalanuvchi ko‘radigan va o‘zaro ta’sir qiladigan sayt qismi. Bu bosqichda dizayn texnik jihatdan hayotga tatbiq etiladi.

Texnologiyalar: HTML (tuzilish), CSS (stil), JavaScript (interaktivlik) asosida ishlanadi.

Frameworklar: Zamonaviy loyihalarda React, Next.js, Vue.js kabi frameworklar qo‘llaniladi. Bu interaktiv va dinamik sahifalar yaratishni osonlashtiradi.

Moslashuvchanlik: Sayt har qanday ekran o‘lchamiga moslashishi uchun responsive yoki adaptive dizayn qo‘llaniladi.

Optimallashtirish: Kodning sifatli yozilishi, brauzerlararo moslik va yuklanish tezligi ta’minlanadi.

💡 Misol: React yordamida foydalanuvchi bir sahifadan ikkinchisiga o‘tmasdan turib interaktiv tajriba olishi mumkin (Single Page Application).

Backend Ishlab Chiqish

Backend — saytning “miyasi”, ya’ni server tomoni. Bu bosqichda saytning funksionalligi va ma’lumotlarni boshqarish tizimi yaratiladi.

Frameworklar: Django (Python), Node.js (JavaScript), FastAPI (Python) kabi texnologiyalar ishlatiladi.

Ma’lumotlar bazasi: Ma’lumotlarni saqlash va boshqarish uchun PostgreSQL, MySQL yoki MongoDB kabi tizimlar qo‘llaniladi.

Funksiyalar:

Foydalanuvchi ro‘yxatdan o‘tishi va tizimga kirishi.

Buyurtma qilish yoki savat tizimi.

Admin panel orqali saytni boshqarish.

API ishlab chiqish: Frontend bilan backend o‘rtasida ma’lumot almashish uchun REST yoki GraphQL API’lar yaratiladi.

Integratsiyalar

Zamonaviy sayt faqat dizayn va funksionallik bilan cheklanmaydi. U turli tashqi xizmatlar bilan integratsiya qilinishi kerak.

To‘lov tizimlari: Payme, Click, Stripe, PayPal kabi tizimlar orqali xavfsiz to‘lovlarni amalga oshirish.

CRM integratsiyasi: Mijozlar bilan muloqotni boshqarish uchun CRM tizimlari (masalan, HubSpot, Salesforce).

Analitika vositalari: Google Analytics, Yandex Metrica yordamida sayt trafigini tahlil qilish.

Ijtimoiy tarmoqlar va boshqa xizmatlar: Telegram bot, email xabarnomalar yoki ijtimoiy tarmoqlar bilan bog‘lanish.

Testlash (QA)

Saytni ishga tushirishdan oldin uni sinchkovlik bilan sinovdan o‘tkazish zarur. Bu xatolarni aniqlash va tuzatish uchun muhim.

Funksional test: Saytdagi barcha tugmalar, formalar va funksiyalar to‘g‘ri ishlashini tekshirish.

Cross-browser testi: Sayt Chrome, Safari, Firefox va boshqa brauzerlarda muammosiz ishlashi kerak.

Mobil test: Mobil qurilmalarda dizayn va funksionallikning to‘g‘ri ishlashi sinovdan o‘tkaziladi.

Tezlik testi: Google PageSpeed Insights yoki Lighthouse yordamida saytning yuklanish tezligi o‘lchanadi.

Ishga Tushirish (Deployment)

Saytni foydalanuvchilarga taqdim etish uchun uni serverga joylashtirish kerak.

Hosting tanlash: VPS (Virtual Private Server), AWS, Vercel, Netlify yoki boshqa platformalar.

Domen ulash: Sayt uchun mos domen nomi tanlanadi va ulanadi (masalan, www.example.com).

SSL sertifikat: Sayt xavfsizligi uchun HTTPS protokoli o‘rnatiladi.

CI/CD jarayoni: Avtomatik yangilanishlar va deploy uchun Continuous Integration/Continuous Deployment tizimlari sozlanadi.

Qo‘llab-quvvatlash va Optimallashtirish

Saytni ishga tushirish loyihaning oxiri emas. U doimiy ravishda yangilanishi va optimallashtirilishi kerak.

SEO optimallashtirish: Kalit so‘zlar, meta-teglar, blog yozuvlari va boshqa usullar yordamida sayt qidiruv tizimlarida yuqori o‘rinlarni egallaydi.

Kontent yangilash: Yangi mahsulotlar, xizmatlar yoki blog postlari qo‘shiladi.

Xavfsizlik: Tizimni yangilash, xavfsizlik yamoqlari qo‘llash va monitoring.

Analitika va yaxshilash: Foydalanuvchi tajribasini yaxshilash uchun tahlillar asosida yangilanishlar kiritiladi.

Xulosa

Veb-sayt yaratish — bu rejalashtirish, dizayn, dasturlash, testlash va doimiy qo‘llab-quvvatlashni o‘z ichiga olgan kompleks jarayon. Har bir bosqich o‘ziga xos muhim ahamiyatga ega va loyiha muvaffaqiyati to‘g‘ri yondashuvga bog‘liq. Agar siz biznesingiz yoki shaxsiy brendingiz uchun professional sayt yaratmoqchi bo‘lsangiz, Innosoft jamoasi sizga eng zamonaviy yechimlarni taklif qiladi.

🔗 Batafsil ma’lumot uchun: Innosoft sayt yaratish xizmatlari

Building an AI-Powered Semantic Search with Django, FAISS, and OpenAI

Azamat Narzulloyev — Sat, 06 Sep 2025 01:34:58 +0000

Semantic search is one of the most exciting areas in modern AI. Instead of matching keywords, semantic search understands meaning, which makes it incredibly useful for e-commerce, chatbots, and knowledge bases.

At Innosoft, we’ve implemented multiple projects where we combine Django (backend), FAISS (vector search), and OpenAI (embeddings) to create high-performance search systems.

In this tutorial, I’ll walk you through:

Why semantic search is better than keyword search

How FAISS works under the hood

How to integrate FAISS + OpenAI with Django

A minimal code example you can try yourself

By the end, you’ll have a working prototype of a semantic search system.

Why Semantic Search?

Traditional search engines rely on keyword matching:

Query: “cheap hotel in Tashkent”

Match: Only documents with exact keywords “cheap”, “hotel”, “Tashkent”.

Semantic search goes beyond keywords:

It understands that “affordable accommodation in Tashkent” has the same meaning.

This is possible using embeddings – vector representations of text.

Tech Stack

Backend: Django (REST API)

Vector Database: FAISS

Embeddings: OpenAI text-embedding-3-large

Language Support: English, Russian, Uzbek

Step 1 – Install Dependencies

pip install django djangorestframework openai faiss-cpu

Step 2 – Generate Embeddings with OpenAI

from openai import OpenAI

client = OpenAI(api_key="YOUR_API_KEY")

def get_embedding(text):
response = client.embeddings.create(
model="text-embedding-3-large",
input=text
)
return response.data[0].embedding

Step 3 – Store Vectors in FAISS

import faiss
import numpy as np

Initialize FAISS index

dimension = 3072 # size of text-embedding-3-large
index = faiss.IndexFlatL2(dimension)

Example data

documents = [
"Best hotels in Tashkent",
"Affordable apartments in Samarkand",
"Luxury resorts in Bukhara"
]

vectors = np.array([get_embedding(doc) for doc in documents])
index.add(vectors)

Step 4 – Semantic Search in Django API

from rest_framework.decorators import api_view
from rest_framework.response import Response

@api_view(["POST"])
def semantic_search(request):
query = request.data.get("query", "")
query_vector = np.array([get_embedding(query)])

distances, indices = index.search(query_vector, k=3)

results = [documents[i] for i in indices[0]]
return Response({"results": results})

Example Query

Input:

{"query": "cheap hotel in Tashkent"}

Output:

{
"results": [
"Best hotels in Tashkent",
"Luxury resorts in Bukhara",
"Affordable apartments in Samarkand"
]
}

Real-World Use Cases

At Innosoft, we applied this approach to:

AI-powered chatbots in the banking sector

Product matching systems for e-commerce

Knowledge bases for enterprises

The result: faster, smarter, and more human-like search experiences.

Conclusion

Semantic search is no longer a luxury – it’s becoming the standard. With just Django + FAISS + OpenAI, you can build scalable, multilingual, and high-performance search systems.

🚀 At Innosoft, we continue pushing the boundaries of AI in real projects. If you’re interested in learning more, check out our work at:
🔗 https://innosoft.uz