Mudit Jain

Founder and CTO at a stealth AI agent startup. Ex Google Deepmind, Ex Head of engineering at Vinbrain (acquired by Nvidia), ACM SIGAI awardee, Menlo VC Fellow.

[Mudit Jain @muditjain87]

Thanks @clattner_llvm for early stage startup guidance yesterday. And congrats on Modular's acquisition by Qualcomm!

[Mudit Jain @muditjain87]

At AI engineers conference in SF this week. This year it seems to be at least 3x bigger than last time.

[Mudit Jain @muditjain87]

@markoa @superplanehq Congrats @markoa and super plane team!

[Mudit Jain @muditjain87]

https://t.co/5wKOMsqSfW - Even better article on this with examples of several verticals. Start with AI automation of outsourced tasks and then expand to insourced ones as models get better.

[Mudit Jain @muditjain87]

AI native services company is a new category where the bet is that higher market cap compensates for lower services margins.
Founders of these companies need to have 3 attributes - love for domain , love for AI agents and love for operations - https://t.co/ucQoX1bvpE

[Mudit Jain @muditjain87]

After talking to a GTM person, I recently realized that "copywrite" and "copyright" are 2 different words and meanings. Till now I used to think it's 1 word copyright with 2 meanings.

[Mudit Jain @muditjain87]

@andrewchen I always follow 2nd approach when giving references.

[Mudit Jain @muditjain87]

@siddsax Good and relatable points. The way to fix it is ask your spouse to get a job and fund you. 😄

[Mudit Jain @muditjain87]

Thank you @DustinABC7 and @abc7newsbayarea for covering and continuing to follow up on mail theft issue in bay area! It was good experience to be interviewed on local news! https://t.co/sID0M0IeFJ

[Mudit Jain @muditjain87]

Saw these humanoid robots at an AI event last week.
Apparently the hands cost as much as the whole robot, and so they are wearing gloves over missing hands.

At $20-40k price range, they'll probably be in every house in future just like a car.

[Mudit Jain @muditjain87]

The latest flex from AI companies seems to be to say at the end of a 15min presentation - "I created these slides on my way to this event using my <AI company> product. Here're the 5 line prompt I used".
Saw this for Manus's presentation few months ago and Perplexity's today.

[Mudit Jain @muditjain87]

@markoa I started setting it up or consider creating a custom one from scratch just this weekend.

[Mudit Jain @muditjain87]

Thanks superplane team @markoa @darkofabijan for hosting the "Agents doing ops" event. Great to meet you in person.

I was looking for a platform like this and excited to see where superplane goes next. 🚀

[SuperPlane @superplanehq]

Full house today at our Agents Unconf event at the House of AI in San Francisco 🙌

We loved meeting so many new people and diving into conversations around different agentic topics.

Thanks to everyone who joined and shared their ideas!

[Mudit Jain @muditjain87]

Great so all founders can just give metric of customer base size/ revenue to a swarm of agents and watch them take it to a unicorn. 😄

[Andrej Karpathy @karpathy]

Three days ago I left autoresearch tuning nanochat for ~2 days on depth=12 model. It found ~20 changes that improved the validation loss. I tested these changes yesterday and all of them were additive and transferred to larger (depth=24) models. Stacking up all of these changes, today I measured that the leaderboard's "Time to GPT-2" drops from 2.02 hours to 1.80 hours (~11% improvement), this will be the new leaderboard entry. So yes, these are real improvements and they make an actual difference. I am mildly surprised that my very first naive attempt already worked this well on top of what I thought was already a fairly manually well-tuned project.

This is a first for me because I am very used to doing the iterative optimization of neural network training manually. You come up with ideas, you implement them, you check if they work (better validation loss), you come up with new ideas based on that, you read some papers for inspiration, etc etc. This is the bread and butter of what I do daily for 2 decades. Seeing the agent do this entire workflow end-to-end and all by itself as it worked through approx. 700 changes autonomously is wild. It really looked at the sequence of results of experiments and used that to plan the next ones. It's not novel, ground-breaking "research" (yet), but all the adjustments are "real", I didn't find them manually previously, and they stack up and actually improved nanochat. Among the bigger things e.g.:

- It noticed an oversight that my parameterless QKnorm didn't have a scaler multiplier attached, so my attention was too diffuse. The agent found multipliers to sharpen it, pointing to future work.
- It found that the Value Embeddings really like regularization and I wasn't applying any (oops).
- It found that my banded attention was too conservative (i forgot to tune it).
- It found that AdamW betas were all messed up.
- It tuned the weight decay schedule.
- It tuned the network initialization.

This is on top of all the tuning I've already done over a good amount of time. The exact commit is here, from this "round 1" of autoresearch. I am going to kick off "round 2", and in parallel I am looking at how multiple agents can collaborate to unlock parallelism.
https://t.co/WAz8aIztKT

All LLM frontier labs will do this. It's the final boss battle. It's a lot more complex at scale of course - you don't just have a single train. py file to tune. But doing it is "just engineering" and it's going to work. You spin up a swarm of agents, you have them collaborate to tune smaller models, you promote the most promising ideas to increasingly larger scales, and humans (optionally) contribute on the edges.

And more generally, *any* metric you care about that is reasonably efficient to evaluate (or that has more efficient proxy metrics such as training a smaller network) can be autoresearched by an agent swarm. It's worth thinking about whether your problem falls into this bucket too.

[Mudit Jain @muditjain87]

@abundand Good point, yeah those jobs will be automated by robots eventually.
This article focuses on AI agents that don't operate in physical domain.

[Mudit Jain @muditjain87]

https://t.co/HMqgoyzF3K this anthropic report says the obvious - jobs that are computer based are most likely to be automated by AI agents eg software engineers, algorithmic traders, accountants, middle managers...

Jobs that are physical are safest eg cooks, farmers, guards etc.

[Mudit Jain @muditjain87]

At Together AI @togethercompute event today. Got to pick brains of people developing FlashAttention and it's newer variants.
I wanted to check if it's possible to scale up to million token/sec anytime in next few years for 8B scale LLMs. As I suspected, answer is most likely no.

[Mudit Jain @muditjain87]

@sundarpichai @Waymo Waiting for waymo technology to be integrated in consumer cars that we can buy!

[Mudit Jain @muditjain87]

Seeing similar things in my personal experience. My work has shifted mainly to high level planning, code reviews and guiding the agent. Much less coding needed.
And most recently the guidance part is also becoming declarative rather than imperitive.

[Andrej Karpathy @karpathy]

A few random notes from claude coding quite a bit last few weeks.

Coding workflow. Given the latest lift in LLM coding capability, like many others I rapidly went from about 80% manual+autocomplete coding and 20% agents in November to 80% agent coding and 20% edits+touchups in December. i.e. I really am mostly programming in English now, a bit sheepishly telling the LLM what code to write... in words. It hurts the ego a bit but the power to operate over software in large "code actions" is just too net useful, especially once you adapt to it, configure it, learn to use it, and wrap your head around what it can and cannot do. This is easily the biggest change to my basic coding workflow in ~2 decades of programming and it happened over the course of a few weeks. I'd expect something similar to be happening to well into double digit percent of engineers out there, while the awareness of it in the general population feels well into low single digit percent.

IDEs/agent swarms/fallability. Both the "no need for IDE anymore" hype and the "agent swarm" hype is imo too much for right now. The models definitely still make mistakes and if you have any code you actually care about I would watch them like a hawk, in a nice large IDE on the side. The mistakes have changed a lot - they are not simple syntax errors anymore, they are subtle conceptual errors that a slightly sloppy, hasty junior dev might do. The most common category is that the models make wrong assumptions on your behalf and just run along with them without checking. They also don't manage their confusion, they don't seek clarifications, they don't surface inconsistencies, they don't present tradeoffs, they don't push back when they should, and they are still a little too sycophantic. Things get better in plan mode, but there is some need for a lightweight inline plan mode. They also really like to overcomplicate code and APIs, they bloat abstractions, they don't clean up dead code after themselves, etc. They will implement an inefficient, bloated, brittle construction over 1000 lines of code and it's up to you to be like "umm couldn't you just do this instead?" and they will be like "of course!" and immediately cut it down to 100 lines. They still sometimes change/remove comments and code they don't like or don't sufficiently understand as side effects, even if it is orthogonal to the task at hand. All of this happens despite a few simple attempts to fix it via instructions in CLAUDE . md. Despite all these issues, it is still a net huge improvement and it's very difficult to imagine going back to manual coding. TLDR everyone has their developing flow, my current is a small few CC sessions on the left in ghostty windows/tabs and an IDE on the right for viewing the code + manual edits.

Tenacity. It's so interesting to watch an agent relentlessly work at something. They never get tired, they never get demoralized, they just keep going and trying things where a person would have given up long ago to fight another day. It's a "feel the AGI" moment to watch it struggle with something for a long time just to come out victorious 30 minutes later. You realize that stamina is a core bottleneck to work and that with LLMs in hand it has been dramatically increased.

Speedups. It's not clear how to measure the "speedup" of LLM assistance. Certainly I feel net way faster at what I was going to do, but the main effect is that I do a lot more than I was going to do because 1) I can code up all kinds of things that just wouldn't have been worth coding before and 2) I can approach code that I couldn't work on before because of knowledge/skill issue. So certainly it's speedup, but it's possibly a lot more an expansion.

Leverage. LLMs are exceptionally good at looping until they meet specific goals and this is where most of the "feel the AGI" magic is to be found. Don't tell it what to do, give it success criteria and watch it go. Get it to write tests first and then pass them. Put it in the loop with a browser MCP. Write the naive algorithm that is very likely correct first, then ask it to optimize it while preserving correctness. Change your approach from imperative to declarative to get the agents looping longer and gain leverage.

Fun. I didn't anticipate that with agents programming feels *more* fun because a lot of the fill in the blanks drudgery is removed and what remains is the creative part. I also feel less blocked/stuck (which is not fun) and I experience a lot more courage because there's almost always a way to work hand in hand with it to make some positive progress. I have seen the opposite sentiment from other people too; LLM coding will split up engineers based on those who primarily liked coding and those who primarily liked building.

Atrophy. I've already noticed that I am slowly starting to atrophy my ability to write code manually. Generation (writing code) and discrimination (reading code) are different capabilities in the brain. Largely due to all the little mostly syntactic details involved in programming, you can review code just fine even if you struggle to write it.

Slopacolypse. I am bracing for 2026 as the year of the slopacolypse across all of github, substack, arxiv, X/instagram, and generally all digital media. We're also going to see a lot more AI hype productivity theater (is that even possible?), on the side of actual, real improvements.

Questions. A few of the questions on my mind:
- What happens to the "10X engineer" - the ratio of productivity between the mean and the max engineer? It's quite possible that this grows *a lot*.
- Armed with LLMs, do generalists increasingly outperform specialists? LLMs are a lot better at fill in the blanks (the micro) than grand strategy (the macro).
- What does LLM coding feel like in the future? Is it like playing StarCraft? Playing Factorio? Playing music?
- How much of society is bottlenecked by digital knowledge work?

TLDR Where does this leave us? LLM agent capabilities (Claude & Codex especially) have crossed some kind of threshold of coherence around December 2025 and caused a phase shift in software engineering and closely related. The intelligence part suddenly feels quite a bit ahead of all the rest of it - integrations (tools, knowledge), the necessity for new organizational workflows, processes, diffusion more generally. 2026 is going to be a high energy year as the industry metabolizes the new capability.

[Mudit Jain @muditjain87]

@dakshgup @grok explain reference to hard seltzer, white claws and other drinks in this blog.