@Rustam-Z⚡️
前往频道在 Telegram
Software Engineer @ Google Book 1:1 session: rustamz.com/calendar I talk about: tech, career, big tech interviews, money. About me: joined Google from 🇺🇿 after university www.rustamz.com linkedin.com/in/rustam-z youtube.com/@rustam-z @rustamz_i
显示更多3 405
订阅者
-424 小时
-17 天
-3330 天
吸引订阅者
七月 '26
七月 '26
+42
在2个频道中
六月 '26
+25
在1个频道中
Get PRO
五月 '26
+113
在6个频道中
Get PRO
四月 '26
+108
在2个频道中
Get PRO
三月 '26
+76
在5个频道中
Get PRO
二月 '26
+108
在4个频道中
Get PRO
一月 '26
+225
在5个频道中
Get PRO
十二月 '25
+59
在1个频道中
Get PRO
十一月 '25
+88
在3个频道中
Get PRO
十月 '25
+179
在8个频道中
Get PRO
九月 '25
+296
在10个频道中
Get PRO
八月 '25
+224
在11个频道中
Get PRO
七月 '25
+64
在2个频道中
Get PRO
六月 '25
+65
在3个频道中
Get PRO
五月 '25
+97
在3个频道中
Get PRO
四月 '25
+134
在2个频道中
Get PRO
三月 '25
+147
在4个频道中
Get PRO
二月 '25
+183
在6个频道中
Get PRO
一月 '25
+375
在4个频道中
Get PRO
十二月 '24
+103
在0个频道中
Get PRO
十一月 '24
+109
在1个频道中
Get PRO
十月 '24
+79
在1个频道中
Get PRO
九月 '24
+141
在5个频道中
Get PRO
八月 '24
+458
在8个频道中
Get PRO
七月 '24
+984
在1个频道中
Get PRO
六月 '240
在3个频道中
Get PRO
五月 '24
+18
在9个频道中
Get PRO
四月 '24
+69
在2个频道中
Get PRO
三月 '240
在2个频道中
Get PRO
二月 '24
+300
在0个频道中
| 日期 | 订阅者增长 | 提及 | 频道 | |
| 28 七月 | +5 | |||
| 27 七月 | 0 | |||
| 26 七月 | +2 | |||
| 25 七月 | +2 | |||
| 24 七月 | +3 | |||
| 23 七月 | 0 | |||
| 22 七月 | +5 | |||
| 21 七月 | 0 | |||
| 20 七月 | +1 | |||
| 19 七月 | +2 | |||
| 18 七月 | 0 | |||
| 17 七月 | +1 | |||
| 16 七月 | +1 | |||
| 15 七月 | +10 | |||
| 14 七月 | 0 | |||
| 13 七月 | 0 | |||
| 12 七月 | 0 | |||
| 11 七月 | 0 | |||
| 10 七月 | +2 | |||
| 09 七月 | 0 | |||
| 08 七月 | +1 | |||
| 07 七月 | +1 | |||
| 06 七月 | +1 | |||
| 05 七月 | +3 | |||
| 04 七月 | 0 | |||
| 03 七月 | +1 | |||
| 02 七月 | +1 | |||
| 01 七月 | 0 |
频道帖子
Google launches Gemini 3.6 Flash and a Mythos rival
Google launches Gemini 3.6 Flash, 3.5 Flash-Lite and security-focused Flash Cyber
🔗 Read more
I hope, 3.5 pro will go public soon
@rustamzokirov
| 2 |  me: can you check my app for security bugs please
fable to FBI:
@rustamzokirov | 1 990 |
| 3 |  Prepare for Yandex ML School 2026 admission
Watch 24 video lectures on math, ML, statistics and algorithms, all in English, to help you get ready before you apply.
What you get:
• core ML theory: from probability to algorithms
• real tasks from last year's selection, broken down step by step
Yandex ML School 2026 is a year-long program that trains aspiring ML engineers through an intensive, offline curriculum taught entirely in English. Applications for 2026 are open now.
🔗 Just register to watch: https://mlschool.yandex.uz
Sponsored content. Disclaimer: Yandex does not provide educational services as part of the Yandex ML School Uzbekistan project.
@rustamzokirov | 2 060 |
| 4 |  when the AI apologizes for the same mistake it made yesterday, with the same confidence
@rustamzokirov | 2 829 |
| 5 |  NotebookLM ➡️ Gemini Notebook | 2 690 |
| 6 |  built an agent for research
she now has strategic opinions on Messi's retirement window 😂
"Argentina — Messi's last real shot at this"
Let's she if she was right
https://t.me/rustamz_workshop/8768 | 2 436 |
| 7 |  How do you build AI agents you can actually trust?
Find out at the AI Agents Security Week, hosted online by Yandex School of Data Analysis (YSDA)!
📅 Date: July 27–31, 2026
📍 Format: free online intensive (5 days of sessions + live Q&A)
🌐 Language: English
Over 5 days, Yandex experts show you the real risks AI agents face and how to protect your projects from the very first design step.
What you will learn:
• understand how autonomous AI agents work and how they differ from LLMs
• how you can protect systems from attacks, data leaks, and other threats
• real security mistakes made by DeepSeek, Microsoft, and Amazon
Who is this for?
This intensive is a good fit for anyone who does a lot of vibe coding and wants to build secure products. To get started, you only need to know how to write prompts and run the generated code.
🔗 Sign up now! Registration is open until July 27, 2026
This program is for informational and educational purposes only to help you build safer AI systems.
Sponsored content. "Y. Izdeu men Jarnama LLP" does not provide formal education services as part of this program. | 2 413 |
| 8 | Here is my future vision for Luna (github.com/Rustam-Z/hamroh)
She will be a companion and an assistant, much like Jarvis from Iron Man.
She should be super proactive in the future, highlighting what is happening that is worth my attention. At the moment, however, the cost of enabling this is too high.
I want her to live in my home and have access to everything: my socials, my email, and the news 24/7.
I want to be able to ask her:
• hey, reschedule the meeting
• hey, message <someone> on Telegram
• hey, ask my friends where they want to go this weekend
• hey, book a table at a restaurant, call them
• she should message me to let me know if someone important has reached out
• she should tell me when I need to buy something for the house
• etc...
This is how I see Luna, and I am sure more and more will be possible in the future.
What is the difference between Luna and OpenClaw or other "claws"? I built Luna. I trust the skills I have, I trust the MCPs I build and connect, and I control my workflows and my tokens. She can improve herself. I am here to direct her and approve things, all from my DMs. | 2 222 |
| 9 |  +5be an engineer
build 3 agents
one runs your life
one runs your startup's team
one watches and fixes the other two
now they talk to each other more than to you
This is the agentic setup I have for my personal use cases, with 3 agents. And I am here to direct them.
The team:
• Nodira (private) — works on startup project, with a team of 15 developers: helps with QA decision making and preparing apps for release, reviews every code change based on team's rules, finds bugs, files them directly from telegram, decides what is urgent, proactive agent, talks with developers directly if needed.
• Mars (private) — the CTO, coder. If I need to update extend capabilities of agents, add new tools, new skills, new MCP, I tell to Mars. It has bash/code access.
• Luna (public) — simple public bot, you all can talk with her at @rustamz_workshop
🔗 And all of them run based on harness framework I have build by myself, framework: github.com/Rustam-Z/hamroh, and here is the agent template: https://github.com/Rustam-Z/hamroh-custom-agent
Go, and try yourself
The coolest feature: I can control everything from my phone, from Telegram. No laptop needed, all agents are deployed in the cloud.
But at the same time, agent that changes itself quietly is not safe to trust. No agent changes its own rules without asking me first. That's not a limit, that's the whole point, just decide the level of independence for your agents.
@rustamzokirov | 2 631 |
| 10 |  Want to actually understand ML under the hood?
🚩 If you want to build a deep structural foundation in AI and ML, applications for Yandex ML School Uzbekistan 2026 are now open
• 1-year
• offline in Tashkent
• fully in English
To help everyone prep for the admissions process, they released a free online prep course:
• 24 video lectures covering math, ML theory, statistics, and algorithms
• solutions and deep dives into last year's selection tasks
• a clear breakdown of what to expect from the entrance exam
Important Deadlines:
• Application Deadline: September 10, 2026
• Online Testing Window: August 27 – September 10, 2026
👉 Apply to Yandex ML School Uzbekistan
Sponsored content. Disclaimer: Yandex does not provide educational services as part of project.
@rustamzokirov | 3 339 |
| 11 |  I have been saying this a lot recently.
Systems thinking and design/architectural thinking, vision, matter a lot now, but they are not easy things to learn. Not everyone can do it.
Only People with "Taste" architect a system that scales, looks elegant, and solves a real human problem. | 2 724 |
| 12 | last one, the one i love 😈
> be a customer
> believe all four above at once
> "AI will save you money"
> "AI will save you time"
> open the monthly bill
> it saved neither
> efficiency achieved
> theirs, not yours | 2 368 |
| 13 | > be Jensen Huang
> wear the jacket
> "AI is the future"
> "everyone must use AI"
> "this will bring trillions in economic impact"
> sell GPUs
> "the more you buy, the more you save"
> nobody's saving except him | 2 239 |
| 14 | > be Demis
> win Nobel Prize
> stay quiet while others panic-tweet
> "we're focused on the science"
> let the other three fight on the timeline
> say that layoffs because of AI are lack of imagination
https://www.wired.com/story/demis-hassabis-ai-layoffs-deepmind-google-io/ | 2 180 |
| 15 | > be Sam Altman
> build chatgpt
> say "we may have just built a digital god"
> tweet AI might crash the economy
> say "I'm delighted to be wrong about this" | 2 075 |
| 16 | > be Dario
> publish 14000 word essay
> scare everyone about superintelligence
> "we might lose control"
> "this could go very badly"
> "50% of entry-level jobs gone"
> "AI will write 100% all the code by end 2025"
> also raise $$$ right after
> do the exact same thing next year
> and the year after
> now claim "automation may actually expand the work people do" | 1 927 |
| 17 | > be Microsoft and block Claude Code for own engineers because it is expensive
> be Uber and spend yearly budget in 4 months with low ROI
> be Starbucks and find out that Al was miscounting inventory
let's see what they are going to do after | 1 890 |
| 18 | what's next:
> be the same company
> panic
> rehire the people you fired
> people now do the work AI couldn't
> AI bill still there
> call it a "strategic realignment" later
> stock goes up anyway | 1 960 |
| 19 | > be a big tech company
> lay off employees to save money using Al
> give Claude Code to engineers to make them "efficient"
> tell everyone to use more tokens
> engineers start using Al
> literally for everything
> code generation
> debugging
> automation
> documentation
> entire workflows
> the monthly bill starts to look like a military budget
> discover that thousands of employees using Al
> all at the same time costs a fortune
> panic | 2 048 |
| 20 | > use your 9 to 5
> to fund your 5 to 9
> to get out of your 9 to 5 | 2 807 |
