Ajinkya Muley

Backend Engineer @Deel - FinTech + AI

[Ajinkya Muley @jynx_29]

New devs really out there learning kubernetes through brainrot videos

[Ajinkya Muley @jynx_29]

Minimax H3 costs about $13 to generate a 90-100 second video.

Kinda insane

[Ajinkya Muley @jynx_29]

https://t.co/q0ZYv2ddjd

[Ajinkya Muley @jynx_29]

5 years ago i was giving an interview where the guy grilled me for not answering a textbook question word-for-word. anyway, later on the call he accidentally opened Vim and literally had to kill the whole terminal because he couldn’t figure out how to exit 😭

[Ajinkya Muley @jynx_29]

New devs really out there learning kubernetes through brainrot videos

[Ajinkya Muley @jynx_29]

5 years ago i was giving an interview where the guy grilled me for not answering a textbook question word-for-word. anyway, later on the call he accidentally opened Vim and literally had to kill the whole terminal because he couldn’t figure out how to exit 😭

[Ajinkya Muley @jynx_29]

@jukan05 So by 2032? Lol

[Ajinkya Muley @jynx_29]

https://t.co/2hR8OM6qFZ

[Ajinkya Muley @jynx_29]

Reason why coding jobs are so competitive now

[Ajinkya Muley @jynx_29]

Work Life Balance

[Ajinkya Muley @jynx_29]

Team: We spent three weeks building a foolproof, intuitive UI.

The end users:-

[Ajinkya Muley @jynx_29]

@jukan05 And knowing AAPL, they would levy more burden on the consumer

Soo like a $500 bump on the 18 Pro Max?

[Ajinkya Muley @jynx_29]

DeepSeek just completely scrambled the AI leaderboards!

Based on the benchmarks and the latest API updates, here is the massive leap they just made:

- Crushes GPT-5.5: Scores 70.3 vs 55.6 on Toolathlon for long-horizon, multi-tool tasks!

- Matches Opus 4.8: Hits 54.4 on DeepSWE, an absolutely insane jump from V4-Pro’s previous 8% score!

- Builds Repos from Scratch: Dominates GLM-5.2 (54.2 vs 48.9) on the NL2Repo benchmark.

- Security Beast: 76.7 on CyberGym for heavy-duty security code analysis.

- Fraction of the Cost: DeepSeek locked in a permanent 75% discount on the V4 Pro API. This undercuts GPT-5.5, Gemini 3.1 Pro, and Claude Opus 4.7 by 3-19x!

Top-tier frontier performance, but way cheaper.

[Ajinkya Muley @jynx_29]

Deepseek V4 flash has just been updated.

What kinda magic did they pull this time? This is so impressive

https://t.co/7YSU32rTLF

[Ajinkya Muley @jynx_29]

Deepseek V4 flash has just been updated.

What kinda magic did they pull this time? This is so impressive

https://t.co/7YSU32rTLF

[Ajinkya Muley @jynx_29]

@jukan05 Roughly on par with Fable 5 with just a fraction of their cost?? 🫣

[Ajinkya Muley @jynx_29]

This is what they mean they say
“This movie was shot on iPhone”

[Ajinkya Muley @jynx_29]

@StockSavvyShay Bought his entire portfolio for pennies on the dollar after spreading rate hike fud

Crazy shit

[Ajinkya Muley @jynx_29]

@vivekst1 Now watch them run it back while most of retail turned bearish haha. Classic

[Ajinkya Muley @jynx_29]

POV: The highest paid engineer at your company shows up for the first time in months

[Ajinkya Muley @jynx_29]

This is absolutely wild 💀

[Ya'el Courtney, PhD @ScienceYael]

ahhhhhhh so I'm hiring a tech to help me out in lab and screening applications now and THE KIDS ARE USING PROMPT INJECTION!!!!!! 2.25 pt white text, here's what I've found so far

[Ajinkya Muley @jynx_29]

Life of developers right until this tweet

[Sam Altman @sama]

today we launched ChatGPT. try talking with it here:

https://t.co/uWra8LKFMN