Dmytro Obodowsky 🇺🇲🇺🇦🇮🇱

If you end up in the environment of high level of expertise in any relevant area in your workplace, consider yourself lucky, you opened yourself to constant learning experience. However, with GenAI you can get an knolageble copilot to yourself in almost any topic. Use it daily.

Once it is done, I will never miss behind the wheel time

Cybercab

@lawyer4SMBs If you are in good physical shape and you don’t hate it, do your chores. Especially if you are white collar worker sitting on your chair all day. Go out mow your lawn, enjoy outdoors

I remember a while back I said that if you make over $200k a a year and still mow your own lawn, you're doing it wrong. Now that the average quote I'm getting for my lawn is $400 a month, I think I'm changing my mind on that statement. What are folks running for lawnmowing equipment on a 1 acre mow these days?

@dev_maims Software Engineer > Coder

Software Engineers, what’s your backup plan if Artificial Intelligence writes better code than you in 2 years?

@thdxr This is the first time imo, when cleaning up tech debt, made convincing execs optional. You see complexity and inefficiency: just spin up some agents and validate the result. Ship it.

are there people out there who just want to refactor every day? just wake up and find the worst code and just chip away at it and clean it up wake up the next day do it again, infinitely improving things with zero external impact?

@javarevisited Dude, stop it. “Only one with the encryption key” is far fetched

A Senior Lead was laid off after 8 years. "It’s just a numbers game," they told him. He handed over his laptop, shook hands, and didn't say a word. Two hours later, the CEO realized that the "silent" developer was the only one with the encryption keys to the production database. The manager called him, expecting him to help for "the sake of the team."↓

@bluewmist Just convince yourself , the only commitment is cross the gym door. The rest is not a commitment. After 10 mins of warmup you are burning on all cylinders

People who exercise even when they don't feel like it, what's your trick?

@shl Depends on codebase

Before, only engineers could contribute to a codebase Now, everyone can

This damn sentiment that we will work less with AI is complete bs. Since I’ve got my hands on almost unlimited token budget , I cannot stop.

Will GenAI eliminate software companies? Unlikely. The "build your own" premise ignores the cost of duplication. Consolidation exists for a reason: shared infrastructure is more efficient than a thousand bespoke clones. Claude is a powerful tool, not a replacement for specialized moats.

As a manager, one of my biggest regrets is losing top performers. They said they we're leaving for a "better opportunity." I figured they left for more money. Turns out, they left because I failed these 3 tests:

@Oblivious9021 One plausible answer would be to separate memory access. Threads can share memory within single process, interprocess memory sharing is possible too but different .

Interviewer : Why do browsers like Google Chrome use seperate Processes for each "tabs" instead of Threads?

@MarcJBrooker Divide and conquer

Here's the technical interview question I'd actually ask. Suppose you have 5 MB of text, how would you count the words? What about 5 GB, TB, PB, EB? When would your approach change and why? What if you had to do it once, or once a week, day, hour, or second?

@0xlelouch_ If messages backlog already sitting in existing partitions, does adding more partitions reshuffle existing backlog across additional partitions, or new partitions utilized ONLY for new messages? If answer is 2nd option, you need to vertically scale your existing workers to catch.

You're using Kafka to stream order events, but consumers are falling behind by 2 hours during peak traffic. How will you catch up? [Production issue at LinkedIn]

@0xlelouch_ If you have huge backlog in this single partition , adding more partitions wouldn’t help. I haven’t worked with Kafka for a long time, but you need to repartition this single one into many and run parallel consumers.

Your Kafka topic has 1 partition, and messages are being processed in order, but throughput is only 100 msg/sec when you need 10K msg/sec. How will you scale this? [Answer post tomorrow]

@brankopetric00 From what I read here your HPA always scales up to the max since 70% is the threshold and with low rps it is still 85%. I recommend to find rps level where your cpu consumption in low 40-ies. Use Fortio for example to apply load to the single node. Also check the Garbage Collecto

HPA is scaling but performance isn't improving. The situation: - HPA target: 70% CPU - Current replicas: 15 (max: 20) - CPU utilization: 85% average - Request latency: still high Observations: - Pods scale up quickly when load increases - Latency doesn't improve with more pods - Individual pod CPU shows 85% even with low request count - Application is CPU-bound computation Pod resources: - requests: 500m CPU, 512Mi memory - limits: 500m CPU, 512Mi memory Node capacity: 4 CPU per node, plenty available. Why isn't horizontal scaling helping?

@SumitM_X Split the traffic between read and writes workloads . Scale them differently and off course read replicas for persistence layer

You are building a microservice that handles read and write operations, but the write operations cause significant delays for read requests. How can you optimize this?

Quite frankly, the most effective solution is to host database in the client. Let me elaborate. “Client” is single in the post, not plural. Client needs all data, 1M users. Assuming effective means, fast and cost effective.

NEVER USE A 401K TO BUY A HOUSE NEVER USE A 401K TO BUY A HOUSE NEVER USE A 401K TO BUY A HOUSE NEVER USE A 401K TO BUY A HOUSE NEVER USE A 401K TO BUY A HOUSE NEVER USE A 401K TO BUY A HOUSE NEVER USE A 401K TO BUY A HOUSE seriously, don’t ever do this 🙂