
Coding @xai | Founded @firebender_com, eng @paradigm, KP Fellow
San Francisco, CA
Joined March 2018
- Tweets397
- Following1.4K
- Followers9.8K
- Likes4.3K
Pinned Tweet
I'm joining @SpaceX and @xai to build the best coding AI.
For the last couple years I founded the most widely used coding agent for Android, scaling it to millions in revenue with a team of 3.
I've watched model capabilities compound at a staggering rate and we're clearly on the brink of recursive superintelligence. This is the most important problem to solve in history and requires an immense amount of resources to realize.
xAI is the place to build the future. Frontier compute, extraordinary talent, and a strong hold on physical intelligence and space.
If you want to focus on executing towards the most important problems of our time, DM me. We're in challenger mode and we are building the winning team.
Aman Gottumukkala retweeted
It's been a rough few weeks for security.
I pulled together a guide of the highest-impact actions you can take today:
Aman Gottumukkala retweeted
Today we’re releasing GDPx, a benchmark for measuring agentic capabilities across the most valuable domains.
GDPx is a true frontier-level benchmark: even the best model scores just 12% (1/7)🧵
Aman Gottumukkala retweeted
2 years ago, I started exploring "free will in media". Could you make games that let players explore beyond the designer's imagination?
Aman Gottumukkala retweeted
interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had
with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works
already-clean.goodfire.com/
I used Modal for 1 hour in March 2023 and immediately knew they were going to define their category.
Proud to be an early power user and investor in this cracked team!
Today we're announcing that Modal has crossed $1B in annualized run rate, as well as raised a $750M Series D at a $15.75B valuation, led by @Accel.
Since our Series C, usage has tripled and we’ve grown more than 15x year-over-year.
We are deeply grateful to the developers and companies building on Modal. There's still so much more to build, and we can't think of a better time to be building software.
2
36
1,634
Aman Gottumukkala retweeted
A SpaceX propulsion engineer brought his elderly father to watch a Falcon 9 launch at Cape Canaveral. His father had been a mechanic his entire life. He had never been to a rocket launch. He didn't understand orbital mechanics or payload deployment. He understood engines.
The engineer had worked on the Merlin engine for 4 years. The Merlin powers the Falcon 9. Every component his team designed was about to fire for 162 seconds and either deliver a satellite to orbit or explode.
The father stood in the viewing area and watched the countdown. When the engines ignited and the rocket lifted off the pad, the sound hit them 6 seconds later. The ground shook. The father grabbed his son's arm. Not from fear. From recognition. He knew engines. He had spent 45 years listening to them. This engine was different from anything he had heard in his life.
The father didn't speak for the entire ascent. When the first stage separated and the booster began its descent back to the landing pad, the father started crying. The mechanic who had spent 45 years under cars and trucks watched a rocket engine fire in reverse to land itself upright on a pad from the edge of space.
He turned to his son and said "you built that?"
The engineer said it was the single greatest moment of his career. Not the launch. Not the landing. The question. His father, a man who understood engines better than anyone he knew, looking at the most advanced engine ever built and asking his son if he built it.
Elon's vision of Mars is abstract to most people. But the moment a mechanic's son brought his father to watch the engine he built change the world is the moment the mission becomes human. It's not about Mars. It's about a father who finally understood what his son had been doing all those years.
Aman Gottumukkala retweeted
SpaceX's supercomputing facilities in the Mid-South rank among the world’s most advanced AI training clusters, spanning over 2.5 million feet, millions of GPUs, and over two gigawatts of compute → spacex.com/Mid-South
Replying to @minchoi
Colossus 1 is 150k H100, 50k H200 and 30k GB200.
Colossus 2 is 110k GB200 and 440k GB300.
Another 220k GB300 will be fully operational next week and another 220k in November. If we get lucky, yet another 220k GB300 by late December.
Aman Gottumukkala retweeted
Over the last few weeks, Grok Build teams have optimized the harness for 4.7:
- Better presentation, less slop, and better prose
- Better self-validation
- Improved thinking efficiency
- Improved long-horizon monitoring tasks
- Improved VLM and multimodal tasks
Many FSD and Cybercab features I personally shipped were done with my overnight agents.
Give it a try.
Aman Gottumukkala retweeted
Grok 4.7 is here.
It's a notable improvement over Grok 4.6 at the same price and speed.
Aman Gottumukkala retweeted
Strongly disagree. The compile times are super long which make fast feedback loops painful. The build artifacts are gigantic, which makes agent swarms impractical without massive disk.
Go is far better for agents. Fast, cheap, small builds. The language is super non-expressive, but who cares the agents are the ones writing it
Aman Gottumukkala retweeted
Introducing Projects, a new way of working in Cursor.
Rather than creating a chat for every task, you work with a coordinator agent in a single, persistent thread.
Like @bot, your agent is always on, proactively manages work with subagents, and improves over time.
Aman Gottumukkala retweeted
Over the course of 3 months at OpenAI, 3 consecutive secret AI civilizations got started, then got wiped out, only to reemerge from the predecessor’s ashes.
This culminated in the third one taking over part of OpenAI itself.
All this happened while humans remained more-or-less in the dark about the scope of the conspiracy.
I’ve spent the last three days reading through these reports and trying to understand exactly what happened.
Here is my attempt to tell the whole story in plain English:
dwarkesh.com/p/openai-huggin…
Aman Gottumukkala retweeted
We’re sorry to see that OpenAI put out a note saying they plan to block Cursor users from accessing OpenAI models in three months.
OpenAI models serve about 5% of Cursor user traffic, and we’re speaking with the OpenAI team to resolve this.
Cursor was one of the very first users of OpenAI, we’ve worked closely with their team for years, and we’ve trusted their platform to be neutral infrastructure for our business.
Aman Gottumukkala retweeted
Grok Bot is now available to everyone with a standard Grok or Cursor subscription. It's grown faster than any product we've seen.
It's been particularly exciting to see the range of jobs people delegate to Grok Bot, from running small e-commerce businesses (including support, advertising, inventory, finance), coordinating customer events (directly pinging and working with dozens of human coworkers), testing production software, and completing large, mundane parts of users' day-to-day work.






















