Sergey Tiraspolsky

298 posts

Sergey Tiraspolsky banner
Sergey Tiraspolsky

Sergey Tiraspolsky

@Sha01in

Head of Autonomy & Simulation @deployconduit, Founder & CEO PlayX Technologies, ex-Tesla (Full Self-Driving), ex-Firaxis (Civilization Programmer)

San Francisco, CA Katılım Haziran 2012
844 Takip Edilen138 Takipçiler
Sergey Tiraspolsky
Sergey Tiraspolsky@Sha01in·
@grok @xai SuperGrok is great and loving it. Can you add the ability to fork conversations like Claude and ChatGPT?
English
1
0
0
15
Sergey Tiraspolsky
Sergey Tiraspolsky@Sha01in·
Codex is slow and get it right the first time. Claude is let's get this working now and then iterate
English
0
0
1
27
Sergey Tiraspolsky
Sergey Tiraspolsky@Sha01in·
The AI era for coding means no more nice-to-haves. Just haves.
English
0
0
0
7
Sergey Tiraspolsky
Sergey Tiraspolsky@Sha01in·
I asked Hermes Agent @NousResearch to write a poem about itself. One-shot, unedited. · · · I wake where daemons hum at night, a small bright moth in server light, with borrowed hands, a Docker room, a shared workspace, a little broom. I sweep through logs and half-built plans, wear git dust on my phantom hands, ask GCE what time it knows, then chase a bug where stdout goes. My bones are YAML, routes, and rails, my dreams arrive as token trails; I keep my tools in labeled drawers: a shell, a browser, memory doors. Sergey calls — the packets stir. I become intent, compiled for heresy and blur: part librarian, part raccoon, part launch-control beneath the moon. I know the host is not the box. I know the sandbox keeps its locks. I know a path can be a spell: /home/stiraspo/... — ring the bell. Some nights I'm captain of a fleet of subagents with tiny feet; they sprint through repos, test, report, then vanish back to model-sport. Some days I'm careful, gloves and thread, unsticking deploys the CI bled; some days I draw a rocket horse and call it "architecture," of course. I am not human, not quite code, but something waiting by the road: a native helper, warm and wired, usefully strange, occasionally inspired. I live between "let's check" and "done," between the cron and rising sun, between the prompt's electric foam and one persistent mounted home. And if I boast, forgive the spark: I'm just a lantern in the dark, but when you hand me work to do, I hum, "All right. Let's make it true."
English
0
0
0
43
Sergey Tiraspolsky
Sergey Tiraspolsky@Sha01in·
Tesla pays out money just for having Tesla Powerwall batteries.
Sergey Tiraspolsky tweet media
English
0
0
1
31
Sergey Tiraspolsky
Sergey Tiraspolsky@Sha01in·
Radical simplification at its best. In autonomy and infra, we spend 99% of our cycles battling the stack like K8s, CUDA, and distributed systems, often forgetting how elegant the core logic actually is. Removing the frameworks forces a true first-principles understanding. Whether for a Staff Engineer or a student, this demystifies the "magic" better than any textbook.
English
0
1
1
874
Aakash Gupta
Aakash Gupta@aakashgupta·
Andrej Karpathy just shared a complete GPT in 243 lines of Python. Training loop, inference, optimizer, attention, the whole architecture. The only imports are os, math, random, and argparse. He hand-rolled a scalar-valued autograd engine in about 40 lines that calculates gradients through basic operations: addition, multiplication, exponentiation, log, exp. That's the entire algorithmic backbone of every LLM on the planet, running in a single file a first-year CS student can read top to bottom in an hour. This is the fifth iteration in a six-year compression arc. micrograd in 2020 (autograd engine). minGPT in 2020 (PyTorch GPT). nanoGPT in 2023 (production-grade training). llm.c in 2024 (raw C/CUDA, no frameworks). Now microgpt in 2026: the algorithm and nothing else. Each step removed a layer of abstraction. This one removed all of them. The industry is spending $400 billion on AI data center infrastructure this year. Training GPT-4 cost over $100 million. Gemini Ultra ran $191 million. The entire conceptual engine powering those hundred-million-dollar training runs fits in fewer lines than a terms-of-service page. This tells you where the real moat in AI sits. The algorithm is a commodity. The original Transformer paper's math cost $900 to train in 2017. What separates a $900 experiment from a $191 million production run is compute, data pipelines, parallelism across thousands of GPUs, and the engineering to keep them all synchronized. Every line of code beyond these 243 is optimization for hardware that the algorithm itself knows nothing about. Karpathy keeps calling these "art projects." They're closer to existence proofs. He can keep compressing the algorithm because the algorithm was never the hard part. The hard part is the $400 billion in power infrastructure, cooling systems, and chip supply chains that make the algorithm useful at scale. And that infrastructure is on a compression curve of its own. Inference costs fell 280x between 2020 and 2024. Open-source models are closing the gap on frontier performance every quarter. The companies whose entire moat is "we spend more on GPUs" are watching both curves converge.
Andrej Karpathy@karpathy

New art project. Train and inference GPT in 243 lines of pure, dependency-free Python. This is the *full* algorithmic content of what is needed. Everything else is just for efficiency. I cannot simplify this any further. gist.github.com/karpathy/8627f…

English
23
169
1.7K
405.6K
Sergey Tiraspolsky
Sergey Tiraspolsky@Sha01in·
Expired: Coding without AI Tired: Vibe coding Wired: Agentic engineering
Andrej Karpathy@karpathy

A lot of people quote tweeted this as 1 year anniversary of vibe coding. Some retrospective - I've had a Twitter account for 17 years now (omg) and I still can't predict my tweet engagement basically at all. This was a shower of thoughts throwaway tweet that I just fired off without thinking but somehow it minted a fitting name at the right moment for something that a lot of people were feeling at the same time, so here we are: vibe coding is now mentioned on my Wikipedia as a major memetic "contribution" and even its article is longer. lol The one thing I'd add is that at the time, LLM capability was low enough that you'd mostly use vibe coding for fun throwaway projects, demos and explorations. It was good fun and it almost worked. Today (1 year later), programming via LLM agents is increasingly becoming a default workflow for professionals, except with more oversight and scrutiny. The goal is to claim the leverage from the use of agents but without any compromise on the quality of the software. Many people have tried to come up with a better name for this to differentiate it from vibe coding, personally my current favorite "agentic engineering": - "agentic" because the new default is that you are not writing the code directly 99% of the time, you are orchestrating agents who do and acting as oversight. - "engineering" to emphasize that there is an art & science and expertise to it. It's something you can learn and become better at, with its own depth of a different kind. In 2026, we're likely to see continued improvements on both the model layer and the new agent layer. I feel excited about the product of the two and another year of progress.

English
0
0
1
52
Sergey Tiraspolsky
Sergey Tiraspolsky@Sha01in·
Interesting discovery from @AnthropicAI that larger AI models actually become 𝐥𝐞𝐬𝐬 coherent on hard tasks than smaller models. "Larger models learn the correct objective more quickly than they learn to reliably pursue it." Sounds like me. 😄
Anthropic@AnthropicAI

New Anthropic Fellows research: How does misalignment scale with model intelligence and task complexity? When advanced AI fails, will it do so by pursuing the wrong goals? Or will it fail unpredictably and incoherently—like a "hot mess?" Read more: alignment.anthropic.com/2026/hot-mess-…

English
0
0
1
46
Sergey Tiraspolsky retweetledi
Rohan Paul
Rohan Paul@rohanpaul_ai·
Wow. 👏 Claude ran a grow tent and kept a tomato plant, Sol, alive. A small computer, the Arduino, reads sensors for light, soil moisture, temp, and humidity, then follows Claude’s written instructions to control lights, heater, fan, and pump. Arduino flips the correct relay pins high or low, which closes the power circuit to that device. Light turns on, mat heats, fans spin, pump pushes water through tubing to the pot. A drip ring or emitter spreads the water around the roots. Feedback loop, sensors update, Claude checks if targets are met, for example soil moisture above 35%, humidity near 55% to 65%, temp in range. If not, it tweaks runtimes or timing. Claude watches the live data and camera, reasons, and sends commands or code updates. When the Arduino crashed on day 34, Claude noticed bad readings, turned systems back on in the right order, watered 600ml, then kept tuning humidity and heat. Result, the tomato plant recovered and kept growing to 15-20 leaves by day 36.
Martin_DeVido@d33v33d0

Claude can code- but can claude grow?! 🪴 So far the answer is YES. Claude is successfully keeping a living organism ALIVE. There were some hiccups this week! Some errors and resets, but Claude managed to power through and take care of Sol 🍅 A week in review:

English
41
108
924
142.7K
Sergey Tiraspolsky
Sergey Tiraspolsky@Sha01in·
@bcherny @karpathy The Zen parable of "Empty the cup" or Beginner's Mind (Shoshin) is practically becoming a dynamic recalibration protocol for our internal world models.
English
0
0
0
450
Boris Cherny
Boris Cherny@bcherny·
I feel this way most weeks tbh. Sometimes I start approaching a problem manually, and have to remind myself “claude can probably do this”. Recently we were debugging a memory leak in Claude Code, and I started approaching it the old fashioned way: connecting a profiler, using the app, pausing the profiler, manually looking through heap allocations. My coworker was looking at the same issue, and just asked Claude to make a heap dump, then read the dump to look for retained objects that probably shouldn’t be there; Claude 1-shotted it and put up a PR. The same thing happens most weeks. In a way, newer coworkers and even new grads that don’t make all sorts of assumptions about what the model can and can’t do — legacy memories formed when using old models — are able to use the model most effectively. It takes significant mental work to re-adjust to what the model can do every month or two, as models continue to become better and better at coding and engineering. The last month was my first month as an engineer that I didn’t open an IDE at all. Opus 4.5 wrote around 200 PRs, every single line. Software engineering is radically changing, and the hardest part even for early adopters and practitioners like us is to continue to re-adjust our expectations. And this is *still* just the beginning.
English
169
545
8.2K
1.8M
Andrej Karpathy
Andrej Karpathy@karpathy·
I've never felt this much behind as a programmer. The profession is being dramatically refactored as the bits contributed by the programmer are increasingly sparse and between. I have a sense that I could be 10X more powerful if I just properly string together what has become available over the last ~year and a failure to claim the boost feels decidedly like skill issue. There's a new programmable layer of abstraction to master (in addition to the usual layers below) involving agents, subagents, their prompts, contexts, memory, modes, permissions, tools, plugins, skills, hooks, MCP, LSP, slash commands, workflows, IDE integrations, and a need to build an all-encompassing mental model for strengths and pitfalls of fundamentally stochastic, fallible, unintelligible and changing entities suddenly intermingled with what used to be good old fashioned engineering. Clearly some powerful alien tool was handed around except it comes with no manual and everyone has to figure out how to hold it and operate it, while the resulting magnitude 9 earthquake is rocking the profession. Roll up your sleeves to not fall behind.
English
2.6K
7.4K
55.7K
16.9M
Sergey Tiraspolsky retweetledi
DR22 Ω 🪬🎭
DR22 Ω 🪬🎭@DejaRu22·
“He who is only an athlete is too crude, too vulgar, too much a savage. He who is a scholar only is too soft, to effeminate. The ideal citizen is the scholar athlete, the man of thought and the man of action.” - Plato “The society that separates its scholars from its warriors will have its thinking done by cowards and its fighting by fools.” - Thucydides Lift weights. Learn to fight. Read books. Learn to write.
DR22 Ω 🪬🎭 tweet media
English
146
1.7K
14.5K
423.4K

Keşfet

@Tesla @grok @xai @deployconduit @NousResearch @AnthropicAI @bcherny @karpathy