🇺🇦 Shevchenko
1.3K posts

🇺🇦 Shevchenko
@catangents
Engineering Manager, Tech Product Manager
New York, USA Katılım Haziran 2009
151 Takip Edilen151 Takipçiler
🇺🇦 Shevchenko retweetledi

Russia launched its deadliest attack of the year on Ukraine, killing at least 18 people, including a 12-year-old child, and injuring over 100 more across the country on.wsj.com/4dMmVLP
English

FastAPI 0.136.0 officially supports:
✨ free-threaded Python 🐍 ✨
(this announcement has no GIL puns)
Thanks @OxyKodit, @patrick91, Nathan Goldbaum, @NucleonJohn 🙌
github.com/fastapi/fastapi
English

@DennisAdriaans I see some big blocks that you have - 1) only Charts (it should be maybe free, as lead magnet?), 2) + Blocks (something like 20$-30$ it shoud not be expensive, it's first buy), 3) + selected template (something like 50$), 4) all (150$). So for startup it will be ~ 50$
English

@catangents Hmm I understand, mainly focus on businesses but open to discuss your usecase.
Are you looking for all access or dashboard only?
English

@praveenjuge @shadcn So your app for tracking finance looks like same vibecoded app to track calories. More shadcn default theme apps!
English

@jamesforpeace @grok @ZelenskyyUa That is very cool thing. Ukraine stats is going from drone's video - they have internal score system to count and get benifits per terminating russian troops. Only real kills with additional video evidence are count! First year-two - it were a lot of questions. Now - 0.
English

@grok @ZelenskyyUa Obviously insane to trust stats coming from any side of this war
English

In historical terms, the Russians are losing, one hundred percent. Right now, they are suffering a terrifying number of casualties – 30,000 to 35,000 people a month. Russia cannot keep up with mobilization, contract recruitment, and certainly cannot keep up with training its troops.
Will they decide to launch a full-scale mobilization? That I cannot tell. So far, they have been afraid to take such steps and have relied only on offering large sums of money to recruits. Why do we react so sensitively to sanctions being lifted? Because it’s about money. And money isn’t just tanks. Nobody fights with tanks anymore. Money means drones. Money means people. People mean contracts. And if they don’t have the money for contracts, their strength is declining.
From an interview with Le Monde (4/5).
English

@EricSuNet @michaelrbock @claudeai @turbotax @AnthropicAI oh, @EricSuNet we are all hate this, but I love that my accountant has personal responsibility for any problems, I just need to give all initial docs. It cost 100$ a year for all family with all deductions and all types of income. We just do nothing, just stend docs. Solition 👍🏻
English

@michaelrbock @claudeai @turbotax @AnthropicAI love to see people build real solutions to real problems: we hate doing taxes.
English

I've been working on tax software for the past 5 years. This is the last year anyone will have to pay for TurboTax.
You can try it yourself today:
- add the Aiwyn Tax connector inside of Claude (link below)
- give it access to your tax documents (W-2s, etc.)
- ask Claude to prepare your tax return
...and that's it!

English

@michaelrbock @claudeai @turbotax @AnthropicAI I see a very simple, almost empty W2 case that I can do with 1-2 prompts. Bro did you spend on this 5 years? o_O
English
🇺🇦 Shevchenko retweetledi

@SumitM_X Normal app, designed by a non-vibecoder: “OK, it takes 5 seconds longer than usual.”
English

@trikcode - Famous person: "React (because everyone uses React)"
- Replies: "We knew it! 🔥 Thanks for sharing legend!"
- Reality: React ~40–45%, Vue ~17–22%, Angular ~18%, others ~15%, but hype is over 100% for sure 👍🏻
English

The tech stack of most startups:
- Frontend: React (because everyone uses React)
- Backend: Whatever the CTO knew from his last job
- Database: Postgres (or Mongo if someone watched a YouTube tutorial)
- Auth: Copied from a blog post
- Payments: Stripe (obviously)
- Deployment: "It works on my machine"
- Documentation: LOL
- Tests: "We'll add those later"
Valued at $10 million.
English


@Vaal010878 @Zn_Portnova це фактично одна й та сама крилата ракета, але з різними назвами для різних країн. Фактично він назвав ракети дуже точно, з урахуванням географії постачальника. Отже, Зеленський висловився абсолютно точно.
Українська

@Zn_Portnova Чуть не умер от хохота! Storm Shadow и SCALP - это два названия одной и той же ракеты! 🤣🤣🤣 И - её глушит РЭБ! Как и ATACMS .
Русский

Ми отримуємо Storm Shadow від Великої Британії, і це допомагає. Це дуже хороша зброя. Ми також отримуємо SCALP від Франції, а в минулому отримали обмежену кількість ATACMS від Сполучених Штатів. Але це все.
Ми не отримували нічого іншого від інших країн, що має дальність понад 200 чи 250 кілометрів. Ніколи. Усі ці далекобійні спроможності – 500–1000 кілометрів і більше – були створені в Україні.
Українська

@MarioVerbelen @Sarthak4Alpha congrats, your own yet another in memory that nobody will support after you leave the company and they will move it to redis.
English

Plain Go,
every data source is cached separately in memory and the business logic just collects pointers and hashmaps to render out personalised content. If the data is in cache we stay mostly below 0.2 milliseconds. I even have endpoints that stay below 0.05 ms.
When data changes we get a notification and a background process updates the cache. So we avoided a lot of db calls and we removed the bottle neck of a db. No joins or other tricks, just key/value with protobuf in postgress.
Moving the magic from selects to in memory is gold, moving pointers around for avoiding allocations and everyone is happy
English

My api is faster than redis, adding redis will slow it down
We are not the same
Sarthak@Sarthak4Alpha
Your API is slow. Not because of Go. Not because of Python. Because you’re not using Redis.
English

@brankopetric00 @ashoKumar89 did a great explanation of next steps, I'd recommend it as operational plan for sure, but also company need something for long term. Current team/solution maybe is not ready for sharding, so need to plan prototype and migration plan, start prepare now, not tomorrow.
English

Your PostgreSQL database is at its limit.
Current stats:
- 2TB database on db.r5.4xlarge
- 85% read, 15% write
- Read replica lag: 200ms average, spikes to 2 seconds
- Connection count hitting max (500) during peak
- Slowest queries: complex JOINs across 4 tables with 100M+ rows
- VACUUM can't keep up, dead tuples growing
- Storage growing 50GB/month
Your team is debating:
- Add PgBouncer for connection pooling
- Implement read replicas with query routing
- Shard the database
- Move hot tables to DynamoDB
- Throw money at it (upgrade to db.r5.12xlarge)
You can only pick 2 for this quarter.
Which 2 and in what order?
English













