📣 Session 2 is live — LLMs & Fine-tuning
Adkham Zokhirov (Google Developer Expert in AI/ML) breaks down how Large Language Models work and how to fine-tune them for your specific use case.

🎥 Recording: https://www.youtube.com/watch?v=yx-J5hg96E4

📁 Session 2 materials:
https://github.com/ai-incubator-org/AI-mentorship-program/tree/main/sessions/Session%202%20-%20LLMs%20%26%20Fine-tuning

🏗 Project — It is a high time to Start Building!
By now you should start working on one of two options:
Option A — Build Ilm AI
A personal AI learning companion based on your own study needs. Full brief shared earlier. (You can keep your repo public if you want to get feedback from others)
Option B — Build your own project
A real product where you can demonstrate everything you learn throughout this program. (even if you choose to work on this option, please refer to Option A requirments so that you know what you need to demonstrate on your own project. You can/should keep your repo private if you do not want anyone to see your code)

More details on how we'll assess both options are coming soon. But here's what to know now:
📋 Weekly demo format (starting next week)
Each week you'll record a short demo covering (we recommend https://www.loom.com/ for 5 minute recordings):
→ What you built that week
→ What challenges you faced and how you overcome those
→ What you learned
✅ By end of next week, you must have:
→ Auth working
→ File upload and processing pipeline running
→ Basic RAG chat answering questions from uploaded documents (covered in 4th session)

⚠️ A few important things to keep in mind:
AI mentorship sessions are just a guide — but they're not enough to equip you with all the knowledge required to build a full app. You'll need to do a lot of self-learning depending on your current level (this program was designed for software engineers with 2+ years of experience, but we welcome everyone to strech their own skills and learn as much as they can).
We still have to cover 2 more sessions needed for the first milestone submission, but we encourage you to start learning and implementing whatever you can now — don't wait.

We recommend 20 hours per week on this project. This is a real product, and the more effort you put in now, the stronger your skills will be by the end of this mentorship program.

Questions? Drop them here in the chat. See you tomorrow at 18:00 Uzbekistan time at Session 3! 🚀

📣 Session 3 is live — Building Production-Ready AI Apps
Ismail Salikhodjaev (AI Engineering Lead) walks through the architecture that makes AI-driven development trustworthy — the harness, context engineering, and the verification layers that close the gap between agent adoption and engineer trust.

🎥 Recording: https://youtu.be/yJRlUTc2sEE

📁 Session 3 materials: https://github.com/ai-incubator-org/AI-mentorship-program/tree/main/sessions/Session%203%20-%20Building%20Production-Ready%20AI%20Apps

🏗 Project — Stop Vibe Coding, Start Engineering
This session is about making sure your project has real structure behind it — not just prompts that sort of work. Use the Ilm AI demo shared in the materials as inspiration to set up your own harness.
That means:
→ Context files, skills, hooks, slash commands — whatever applies to your project
→ Although we use Claude Code as a default, you are free to use free tools Ismail mentioned during the session — the concept is still the same
→ The goal from this session is that your project is engineered, not just vibed into existence. You should be able to trust the output it produces.

If you're building Option A (Ilm AI) or Option B (your own idea), the expectation is the same: it is advised that by Tuesday, there should be an engineering setup behind your code plus the auth part of the project with different user roles like admin, regular user, etc.

Questions? Drop them here in the chat. See you at Session 4 on Tuesday! 🚀

👋 Hey everyone!

It’s been an incredibly productive and energizing week here at AI Incubator. A huge thank you to everyone who has shared their stories with us so far — this is exactly the kind of community and energy we want to build together.
Now it’s our turn.

Meet the founders behind AI Incubator — two different journeys united by one mission: bringing together strong tech talent and business expertise to help founders from Uzbekistan build real, scalable startups.

✨ Meet Rano Sidikova
Rano previously served as a Partner Success Manager at Plug and Play Tech Center Uzbekistan — a global innovation platform and VC firm connecting startups with corporates, investors, and universities worldwide. She has also mentored and served as a jury member at Pizza Pitch and numerous startup hackathons across the region.
Over the years, Rano has designed and led startup and community programs across Central Asia. She managed two Tumaris Tech initiatives supported by IT Park Uzbekistan and USAID’s Future Growth Initiative, helping more than 40 women founders from Central Asia scale their startups internationally.
As a Google Women Techmakers Ambassador, she has organized and led tech events for over 1,000 participants across Uzbekistan, collaborating with communities and institutions such as School 21, Najot Ta’lim, PDP Academy, Mohirdev, and the Muhammad al-Khwarizmi Specialized IT School.

She is deeply passionate about mentoring young talent and creating opportunities for future founders. One example is Digital Valley Camp, where 60 IT students from Uzbekistan, Kyrgyzstan, and Tajikistan learned how to transform ideas into startups.

🏆 Her contributions have been recognized with:
• Yulduz Tech Award by the Ministry of Digital Technologies of Uzbekistan (2024)
• Kursiv Media’s Best Among Equals Award in ICT (2024)
• TechWomen Seed Grant for Impact Project in Silicon Valley (2024)

Rano holds a Bachelor’s degree in Business Administration from Westminster International University in Tashkent.

🔗 Connect with Rano on LinkedIn:
https://www.linkedin.com/in/rano-sidikova/


✨ Meet Munira Begmuratova
Munira is a Senior Software Engineer at a US-based startup, and a member and community leader at Toptal — the global platform for the top 3% of technical talent.

Her path into tech wasn't a straight line. She started in academia, teaching Finance, Business, and Math at Westminster International University in Tashkent, then coordinated WIUT's Center for Entrepreneurship. After attending the TechWomen program in the US in 2016, she made the switch from finance to IT in 2017, relearning coding from scratch. In 2019, she joined EPAM as a software engineer and volunteered as Country Coordinator at The Rolling Scopes School, the free online coding bootcamp backed by EPAM.

She holds a Bachelor's in Business Computing from WIUT and a Master's in Entrepreneurship & New Venture Creation from Lund University in Sweden.

🏆 Along the way, she was awarded the CEO People Developer Award at EPAM (2020), the Yulduz Tech Award from the Ministry of Digital Technologies of Uzbekistan (2024), and the UK Alumni Award in Science and Sustainability (2025).

Passionate about empowering others in STEM, she frequently speaks at local and international events on career development, AI, entrepreneurship, and programming, and actively mentors and judges at hackathons.

After watching startups from both sides — the business side and the tech side — she wanted to put the two together, inspired by programs like nFactorial Incubator in Almaty and others.

🔗 Connect with Munira on LinkedIn: https://www.linkedin.com/in/mbegmuratova/

Together, we are building AI Incubator to do one thing well: take serious tech talent and serious business experts, and put them in the same room to build real startups.

Let's build future AI startups together 🚀

📣 Finally Project Submission Requirements — How to Share Your Progress

Every participant builds an AI app individually and documents the build publicly in github. Here's what that looks like:

🏗 Pick your project (and stick with it)
→ Option A — Build Ilm AI: brief and milestones are already defined, just start building if you haven't yet started yet https://docs.google.com/document/d/12Tn0PtOv4hkg_lhAlP7xuzT2yNsxUScDIjwqjRIXl-Q/edit?usp=sharing
→ Option B — Your own startup idea: same rules, but you scope your own features based on the milestones shared in https://github.com/ai-incubator-org/AI-mentorship-program/blob/main/README.md#project-milestones
Your product may not have a "quiz mode" or a "Telegram bot", but it should hit the equivalent stage at each point based on what you learnt from your mentors during the sessions shared in youtube https://www.youtube.com/watch?v=yJRlUTc2sEE&list=PLUkXcMtliKco2bUjG73YBmd_VzpH8XqnQ

📁 One GitHub URL for the whole program
→ Public repo with a diary/ folder inside — preferred
→ If your startup code is private, create a separate public diary repo

Either way, the diary must be public — peers and mentors need to read it to review your work.

📋 Diary entries (starting Week 2)
→ At least 2 entries per week in diary/YYYY-MM-DD.md format
→ Each entry covers: what you did, problems you hit, solutions, next plan, time spent

🎥 Loom videos
→ Max 5 min, up to 2 per week
→ Embed the link in your diary entry under ## Demo video

Full project submission requirements are here:
https://github.com/ai-incubator-org/AI-mentorship-program/blob/main/REQUIREMENTS.md

We'll share detailed instructions on where to submit by the end of this week hopefully. Thank you!

Questions? Drop them right here in the chat 🚀

📣 Session 4 is live — Retrieval-Augmented Generation

Arman Grigoryan (Senior ML Engineer) walks through how to build RAG systems that actually hold up on real company data — treating retrieval as a product system, not just prompt engineering.

🎥 Recording: https://youtu.be/2Rg7b6CssqM

📁 Session 4 materials:
https://github.com/ai-incubator-org/AI-mentorship-program/tree/main/sessions/Session%204%20-%20RAG

🏗 Project — time to wire up your RAG pipeline
This session is the backbone of your first milestone submission, the document-Q&A part of your project.

✅ By the end of this week, you must have:
→ Auth working
→ File upload and processing pipeline running
→ Basic RAG chat answering questions from uploaded documents

Don't over-engineer it — get the baseline working end to end first. Hybrid search, reranking, and eval can come later once the simple version answers work correctly. As Arman put it in the session: start with RAG + evals, and only add complexity when the product actually demands it.

⚠️ Important: the session time has changed from 18:00 to 20:00 Uzbekistan time, as requested by the mentor!

Questions? Drop them right here in the chat. See you later today 20:00 Uzbekistan time at Session 5! 🚀

📣 Session 6 is live — Orchestration Frameworks

Ismail Salikhodjaev (AI Engineering Lead) breaks down how to orchestrate reliable multi-agent systems in production — the four canonical patterns, why you ship the graph and not the agent, and how to build it with LangChain + LangGraph.

🎥 Recording: https://www.youtube.com/watch?v=g9yJ8v_m4ew

📁 Session 6 materials:
https://github.com/ai-incubator-org/AI-mentorship-program/tree/main/sessions/Session%206%20-%20Orchestration%20Frameworks

⚠️ Remember: the session is just a guide, a very good guide, but please do the self-learning to fill the gaps.

📣 Session 7 is live — External Integrations
Daz Akbarov (Solutions Architect at AWS) breaks down how agents reach the outside world — APIs, databases, and webhooks — and how to wire them up so they're reliable and safe in production.

🎥 Recording: https://youtu.be/w8qxcpN1k7w

📁 Session 7 materials:
https://github.com/ai-incubator-org/AI-mentorship-program/tree/main/sessions/Session%207%20-%20External%20Integrations

🏗 Project — time to plug your app into the real world
This is the session that turns your project from a chatbot into something that actually does things. By now your agent should be able to reach out and act, not just talk. This week, get at least one real external integration working end to end:
→ Pick one integration that fits your product — a Telegram bot, a payment flow, or file storage
→ Add proper auth (least-privilege — never put user secrets in the prompt)
→ Make the call reliable: set a timeout, add retries, use an idempotency key for any write

⚠️ A couple of things to keep in mind:
Treat anything coming back from a tool or webhook as untrusted input — prompt injection via tool output is the #1 risk on the OWASP LLM Top 10.

👋 Hey everyone!

Quick update — yesterday we set the deadline for our first milestone, and I've now shuffled everyone into peer-review groups. 🎉

🎥 Please watch this short Loom for how it works:
https://www.loom.com/share/bf4087af14f940cb95f0dd05d58e9b9e

Here's the gist:
→ Everyone who submitted gets 4 diaries to review
→ Once you've done your reviews, you'll be able to see your own mark — as long as at least one of your reviewers has marked your diary too

So the sooner we all review, the sooner everyone gets their feedback. 🙌

One small ask: please be kind and generous with your feedback. A thoughtful, constructive note helps your peers way more than a quick "looks good" — and you'll get the same care back on your own work.

Thank you all, and happy reviewing! 💛

📣 Session 8 is live — Generative Computer Vision

Arman Grigoryan (Senior ML Engineer at EPAM Systems and lecturer at IT Park University) walks through generative computer vision — how models create, edit, and simulate visual content, and how to turn that into a focused startup MVP. He covers the core model families (VAEs, GANs, diffusion), why this is a real startup opportunity right now, and a hands-on text-to-image and image → analysis → prompt → image workflow in Python — the demo code is in the GitHub repo!

🎥 Recording: https://youtu.be/sihaTUzI1Lo

📁 Session 8 materials:
https://github.com/ai-incubator-org/AI-mentorship-program/tree/main/sessions/Session%208-Generative%20Models

🏗 Project — we're at the halfway mark
That's Session 8 of 16 done, which means you're officially halfway through the program.

Whether you picked Option A (Ilm AI) or Option B (your own idea), by the end of this week you should have:
→ Quiz mode live
→ Learning plan agent generating plans
→ Telegram bot sending reminders and running quizzes

Option B — aim for something of similar weight on your own startup idea.

And don't forget your diaries and a Loom video walkthrough of what you've built and how you got there!

Questions? Drop them right here in the chat. 🚀

📣 Peer reviews for the second milestone are now open!
If you submitted your GitHub repo, you'll receive up to 4 peer reviews from other participants.

⚠️ One important thing to keep in mind:
You'll only see your own mark once you've reviewed other participants' work too.

We hope you learn from reading your peers' code along the way.

You can also leave your feedback as you watch Loom videos of your peers!

Questions? Drop them right here in the chat. Good luck everyone! 🚀

📣 Session 9 is live — Multimodal AI Startups
Arman Grigoryan (Senior ML Engineer at EPAM Systems and Lecturer at IT Park University) walks through how to build AI-powered products that combine text, images, audio, video, and documents — from CLIP and VLMs to unit economics and shipping your MVP.

🎥 Recording: https://youtu.be/ZYZnBmB0b50

📁 Session 9 materials:
https://github.com/ai-incubator-org/AI-mentorship-program/tree/main/sessions/Session%209%20-%20Multimodal%20AI

🏗 Project — Week 3 milestone
By now you should have:

✅
→ Knowledge gap detection running across sessions
→ Payment flow integrated in test mode
→ UI complete and mobile-friendly

This is a significant milestone — your product should be starting to feel like a real app at this point. If any of these feel far away, prioritize getting a rough version working over making it perfect.

Questions? Drop them right here in the chat.

📣 Session 10 is live — ML, AI & Gen AI
Daz Akbarov (Solutions Architect at AWS) walks through the full AI stack — from classical ML paradigms and algorithms through deep learning architectures to Foundation Models and Gen AI — and shows how each layer builds on the one before it.

🎥 Recording: https://youtu.be/o6FQBFmRRMk

📁 Session 10 materials:
https://github.com/ai-incubator-org/AI-mentorship-program/tree/main/sessions/Session%2010%20-%20AI%2C%20ML%20and%20GenAI

See you at Session 11 today! 🚀

Manashu mavzuga mos o'zimdagi, ishxonadagi tasklarni qilishdagi setupimni qisqacha ko'rsatmoqchiman.

Bu yerda ham loop injinering bor. Faqat 100% emas. Yuqorida chatoq degan joylar uchun inson aralashuvini qo'shganman.

Menga manashu ishni qilib ber deb task berishsa, manashu rasmdagidek jarayonlar bosib o'tiladi. Men shunchaki Yes Yes Yes bosib o'tirmayman, 1.6 o'ynayman :D (hazil)