Dhruv Kaushik

1.1K posts

Dhruv Kaushik

@drvxksh

Chasing my ghost

Katılım Temmuz 2020

90 Takip Edilen60 Takipçiler

Dhruv Kaushik@drvxksh·1h

Atleast someone accepted it

Dexerto@Dexerto

Xbox CEO Asha Sharma says they need to "deepen their connection" with the community and retire features that don't align with their future plans "We will begin winding down Copilot on mobile and will stop development of Copilot on console"

English

Dhruv Kaushik@drvxksh·2h

@zeddotdev The biggest thing i keep going back to vs code is to resolve merge conflicts. Hoping to see it soon

English

Zed@zeddotdev·15h

Code lens support lands in stable .... tomorrow! 🎉

English

1.5K

65.6K

Dhruv Kaushik@drvxksh·11h

Miranda lowkey reminds me how important it is to maintain your calm and composure even in the craziest storms. An art i believe to not let your inner emotions be reflected on the outside

English

Dhruv Kaushik@drvxksh·5d

@ojasgoyalx Maybe I'll do it now

English

Ojas Goyal@ojasgoyalx·6d

@drvxksh Well you would know of it , if you had done CS50 Python

English

Dhruv Kaushik@drvxksh·28 Nis

Came across the "walrus" operator while fiddling with a library today. I always thought its a golang thing. Never expected to see this in Python !

English

Dhruv Kaushik@drvxksh·5d

@zeddotdev So happy its finally here. I've bee using this from the earlier days, even before they had edit predictions. It's been a long journey. Really proud of the team 👏

English

119

Zed@zeddotdev·6d

Zed 1.0: Your last next editor.

English

120

300

470.6K

Dhruv Kaushik@drvxksh·12 Nis

Is it normal for ChatGPT to sprinkle hindi here and there?

English

Dhruv Kaushik@drvxksh·1 Nis

RIP my childhood where i would create new accounts to have a better Gmail address. Never expected this to drop before GTA 6. 2026 is wild

Sundar Pichai@sundarpichai

2004 was a good year, but your Gmail address doesn't need to be stuck in it. To say goodbye to v0t3f0rp3dr02004 @gmail.com or mrbrightside416@gmail.com (or whatever you were into at the time), go to your Google Account settings and choose any name available. You'll keep your old username and you can sign in with both.

English

Dhruv Kaushik@drvxksh·1 Nis

It's like a full circle moment when I actually use these concepts at work. Last month I wrote a k-way merge sort to sort a huge CSV file. All those steps and logic make so much sense when you actually implement them.

Dhirendra@cmd_dj221

“DSA is useless in real development.” True, until you have to build something real. This mini auto-suggest engine I built recently is a live example (few more are there from my early days). It wasn’t about calling an API or plugging in a library. It forced me to think: • How do I return results in milliseconds? • How do I scale prefix search efficiently? That’s where a Trie came in. No abstraction, No shortcut, Just raw problem solving and that’s the part people miss. Most of the time, you will work with abstractions, frameworks, etc, but the moment you start building systems instead of just features, DSA starts to matter. Not all of it, but enough to: • Design efficient structures • Optimize performance • Build your own “engines” when needed DSA isn’t your daily tool, frameworks already use it under the hood, so you don’t have to, but it’s your edge-case superpower. Even I don’t know or remember a lot of DSA: • Do I know everything about Tries? No • Can I solve every Trie problem? No • Can I solve every LeetCode problem? Definitely no But can I learn and use data structures when required? Hell yes. And that’s what matters. Don’t do DSA just for interviews, Do it so that when the moment comes, you’re the most efficient engineer in the room, going beyond frameworks. I didn't have this kind of awareness early in my career, due to which i lost many years just roaming around abstractions and lost many good opportunities. If you’re early in your journey, don’t make that mistake.

English

Dhruv Kaushik@drvxksh·29 Mar

@flipkartsupport You've always been there only to give me updates that I already know. I've shared the details in the DM. Let's see if you're any different

English

FlipkartSupport@flipkartsupport·29 Mar

@drvxksh I'm extremely sorry that we've kept you waiting for your order. We're always here to help you out! Please share the order details with us via DM so that your details remain safe on the social platform. Awaiting your response. twitter.com/messages/compo…

English

Dhruv Kaushik@drvxksh·29 Mar

@Flipkart I cannot tell you how many problems there are with this. 1. You kept playing with the package from Thane to Mumbai and delayed the delivery (don't give me that warehouse bullshit, I have even checked that on ekart, it's literally same) 2. You fail the delivery for god knows what unexpected reason without any call. 3. It's been 3 days now since that failed delivery, did you retry it? 4. Been contacting @flipkartsupport and they just keep telling me wait until this deadline. You've shifted that deadline twice goddammit. You guys have nothing to loose, you'll keep extending the deadline or just give me a refund but what is with this customer experience? How will you compensate for my time? What happens with this customer is that he either gives up and asks a refund, or he goes to the warehouse and picks it himself. Do you realise how shitty your logistics are? You have one job and that too the customer is doing for himself

English

Dhruv Kaushik@drvxksh·27 Mar

@wojakcodes Interesting take.

English

160

‎Wojak Codes@wojakcodes·27 Mar

and make it cold

GRITCULT@GRITCULT

Take a shower in the dark.

English

241

Dhruv Kaushik@drvxksh·17 Mar

@git_cuber Its sounds so scary now that you illustrate this

English

Arpit Chaudhary@git_cuber·17 Mar

🚨 Your future boss just got promoted. And it’s not human. OpenAI just quietly crossed a line. AI agents can now create other AI agents. Not a developer. Not a company. The agent itself. While you sleep. Imagine this: • One AI gets a task • It spins up 10 specialized agents • One researches • One writes • One codes • One analyzes data • One checks quality Then it manages them like a team lead. No human in the loop. This is the real shift nobody is talking about: AI isn’t just replacing workers anymore. It's building org charts. An AI manager… hiring AI employees… that can hire their own AI employees. So the competition isn't: You vs AI It's: You vs an AI company of 50 agents. Running 24/7. Never tired. Never waiting for meetings. You're not being replaced by AI. You're being replaced by AI’s employee. And now… AI’s employee just got employees of its own. Welcome to the agent economy. (And most people still think ChatGPT is just a chatbot.)

OpenAI Developers@OpenAIDevs

Subagents are now available in Codex. You can accelerate your workflow by spinning up specialized agents to: • Keep your main context window clean • Tackle different parts of a task in parallel • Steer individual agents as work unfolds

English

389

Dhruv Kaushik@drvxksh·13 Mar

@arpit_bhayani How do you build that 5/5? When the work isn't in one particular niche and the tools keep changing, how does one decide what's the 5/5 he wants to develop?

English

Arpit Bhayani@arpit_bhayani·12 Mar

I feel the tech industry will have spiky teams in the future, i.e., teams with candidates who are 5/5 in some areas rather than 3/5 across everything. A candidate who holds 5/5 in databases and 2/5 in low-level design is more valuable than someone who is a solid 3/5 across the board. The 3/5 person seems to be a safe bet, but it would be nearly impossible to drive outsized impact with a fully balanced individual. The thesis is simple - extreme strengths are rare and hard to develop. A 2/5 weakness can often be coached up to a 3/5 with time. But you cannot coach someone into being exceptional at something. So, as the world becomes more aggressive and cutthroat, well-rounded candidates will create an impedance mismatch. They fit everywhere but excel nowhere. A team of 3/5 generalists will be competent (no doubt), but highly unlikely to be remarkable. The better bet would be to hire people with sharp edges, pair them with teammates whose spikes complement their gaps, and coach up the soft spots over time to become average. Exceptional ones have never looked balanced on paper.

English

957

80.7K

Dhruv Kaushik@drvxksh·11 Mar

So I was working on clustering geospatial objects (like traffic signals/signs) and I had to revisit DBSCAN vs HDBSCAN. In my observation DBSCAN is the right choice when your clusters are defined, your requirement is strict and you aggressively want to remove outliers. HDBSCAN on the other had makes sense for a use case like me where the density varies and stability is more important than the density. This is when all that "robustness" that they used to talk about in college made sense to me.

English

Dhruv Kaushik@drvxksh·10 Mar

@zeddotdev Yayyy, so happy it came before I passed out of university ❤️

English

126

Zed@zeddotdev·10 Mar

Introducing: Zed for Students! 🎓 Enjoy Zed's Pro plan free, for a year, if you're a current university student (or teacher!) - Zed Pro features for 12 months - $10/month in token credits - Unlimited edit predictions Apply today: zed.dev/education

English

234

396

4.8K

595.5K

Dhruv Kaushik@drvxksh·5 Mar

I canceled my Claude subscription this month, and it feels really good, in my opinion. I feel more alive than doing that clerical job of struggling with the AI to get things right.