←
Ainesh Chatterjee5,925 posts
Ainesh Chatterjee avatar

Ainesh Chatterjee

@ain3sh

Agent Relations @Factory I code often and sleep occasionally. I like systems. Agents are graphs. Harness is everything. Views are mine. #2 fan of Corvus Corax

◷ Joined November 2023▧ 228 Photos and videos
Ainesh Chatterjee avatar
Ainesh Chatterjee@ain3sh·Oct 29

Harness matters. Since the days of TB2, other offerings in the space are now much more considerate with their harness' design, but experience, relentlessness, grit, and ofc principled research, works wonders!

Introducing Droid 1.0: the new engine behind Factory’s agent. Built for complex projects and everyday development: >more accurate >more efficient >faster than ever before

0:18
Wevolver avatar
⟳ Ainesh Chatterjee retweeted
Wevolver@WevolverApp·Nov 17

Researchers have used reinforcement learning to teach a 5-tonne John Deere skid-steer loader to perform complex acrobatic manoeuvres autonomously. Trained in simulation and transferred to a real machine, the system can balance on two wheels, navigate rough terrain, spin, and control its bucket before landing safely. The research demonstrates the potential of AI-driven automation for heavy construction and industrial machinery. Video Credit: @leggedrobotics / @ETH Paper: wevlv.co/4kltYys #engineering #technology

1:10
Minh Nhat Nguyen 🦭 avatar
⟳ Ainesh Chatterjee retweeted
Minh Nhat Nguyen 🦭@minhnguyen·20h

>"jensen is just another CEO" >TSMC 1 week after you remove Morris Chang's drinking buddy of the past 30 years:

TSMC now wants the cash up front. With so much money moving in circles between chipmakers, cloud providers and AI labs, you can see why. "Regarding orders, orders exceeding the credit limit are paid in cash, but some customers require upfront payment to secure designated production capacity." news.mydrivers.com/1/1166/11…

Post attachment
Factory avatar
⟳ Ainesh Chatterjee retweeted
Factory@factory·11h

NVIDIA post-trained Nemotron with Factory, running entirely within its own air-gapped network. Factory was the only commercially supported harness deployed by @nvidia in their environment, giving researchers access to long-running agents and full execution traces.

5:02
geoff avatar
⟳ Ainesh Chatterjee retweeted
geoff@GeoffreyHuntley·Nov 17

taps sign

Property-based tests generated from the spec and from the code catch different bugs. Use both. In a new paper, across 25 agent-built apps, Spec-Driven PBT found 56 bugs the Code-Driven tests missed, and Code-Driven PBT found 43 that Spec-Driven missed, so 59% of bugs were found by only one approach. Setup - An agent writes invariants in plain English from the requirements doc - Then it reads the code to turn them into property tests - The baseline derives properties from the code alone Why PBT? - Each property-based test kills about 50× as many mutants as the average unit test (prior work cited in paper) - Properties double as documentation, far shorter than the implementation Bugs only one approach found, by type - Spec-Driven: mostly state, lifecycle and logic errors (61% of 56) - Code-Driven: mostly input handling (49% of 43) Limitations - Bugs were judged by the same model that wrote the tests (Claude Opus 5); 25 sampled by hand, all held up - The prompts differed between approaches - The method needs a good spec Insight "Invariants obtained from the specification are not biased by the behaviour of a particular implementation, and can therefore falsify behaviour about which the implementation is internally consistent." My take The code can’t be the only oracle, the spec must be one too. Next: move key properties into types and checks, so they hold by construction. arxiv.org/abs/2611.15920

Post attachment
Eva Hill avatar
⟳ Ainesh Chatterjee retweeted
Eva Hill@EvaXYZH·12h

Just one more 2FA step bro. I promise bro just one more 2FA step and it’ll fix everything bro. Bro. Just one more 2FA step. Please just one more. One more auth layer and we can fix this whole problem bro. bro cmon just give me one more 2FA step i promise bro

will brown avatar
⟳ Ainesh Chatterjee retweeted
will brown@willcb·Nov 17

the problem with markdown is that agents write it as data and read it as code

Factory avatar
⟳ Ainesh Chatterjee retweeted
Factory@factory·Nov 17

Introducing Chained Automations. You can now: • Chain multiple automations together, with one triggering the next • Pass automation outputs as context to downstream runs • Trigger automations from schedules, Slacks, webhooks, GitHub events, & more

0:26
dex avatar
⟳ Ainesh Chatterjee retweeted
dex@dexhorthy·Nov 17

my kingdom for a model that credibly and reliably says "are you sure you want X? It would be much simpler to Y"

Peter Gostev avatar
⟳ Ainesh Chatterjee retweeted
Peter Gostev@petergostev·Nov 17

I'm really struggling with this. Not saying that he is wrong on everything, but listening to the scenarios feels like playing a game with a child, the rules of which he is inventing as we go, to end up with a pre-determined outcome. -- Why would AIs take over the labs? -- They would want to control everything to cheat on graders -- How would the AIs take over wider control? -- They wouldn't want anyone finding out that they cheated so they would start hacking governments and taking control over resources -- What if government will show up with guns to stop it? -- They would control the public opinion against that -- How would they control public opinion? -- They would hack into all devices and control information flow -- What if we manage to stop the big data centres anyway? -- The models will get small enough and will be spread everywhere to control everything Again, I don't want to say that we don't need to take this seriously - his main point of improving security seems unambiguously like a good thing. But the idea that we are on a realistic path towards AI's taking over the world in a year just seems like a fantastical idea that is derived by starting from that as an outcome and then backing into it with ever more esoteric unlikely scenarios that must all happen together for this to occur.

.@paulfchristiano thinks that on the current trajectory (absent any slowing down), there is a 1 in 3 chance that we could have superintelligence powerful enough to overthrow the government within a year. (He articulates a concrete scenario). He argues that the labs should pause further algorithmic progress until they can secure their weights and research secrets. Otherwise China could simply steal the superintelligence in the middle of the intelligence explosion, and the labs would no longer be able to slow down even if they saw their models going off the rails. He also explains how AIs will be trained during the intelligence explosion, and why even a system trained entirely inside a data center could still have the capabilities profile such that humans would have very little chance of competing with it or controlling it. And he discusses the plan for alignment - basically, making a superintelligence that actually does what we ask the way we mean it. Fixing reward hacks and building better graders will help up till the next few model generations, but he thinks that no single current approach has better than even odds of letting us control superintelligence. The hope is that the AIs figure out the solution to this problem during the intelligence explosion. Paul is head of the Alignment Research Center. Previously, he was head of AI safety at the US AI Safety Institute. Before that, he led alignment at OpenAI, where he invented RLHF. He’s recently joined OpenAI’s board (though here he is speaking as a private citizen).

1:38:45
AI.Engineer avatar
⟳ Ainesh Chatterjee retweeted
AI.Engineer@aiDotEngineer·Nov 17

"If you want good taste, you have to eat." @jxnlco of @OpenAI on studying good work, trying other people's tools, and learning to explain what makes something good. From AI Engineer Code Summit.

1:35
Jack Morris avatar
⟳ Ainesh Chatterjee retweeted
Jack Morris@jxmnop·Nov 17

dipping my toe into the formal verification space

Post attachment
Rhys avatar
⟳ Ainesh Chatterjee retweeted
Rhys@RhysSullivan·Nov 16

your job as a software engineer in 2026 is to scroll twitter for useful things to send to your agent

you can now test Artifacts repos locally! - create, fork & import repos with wrangler dev - clone & push without impacting what's in production - repo events (repo.forked) can trigger your workflows locally, so you can test them before you deploy big shoutout to @leostera for shipping this!

Post attachment
dex avatar
⟳ Ainesh Chatterjee retweeted
dex@dexhorthy·Nov 17

context engineering is software research

Why I am sick of the status games around researchers in San Francisco “I am so sick of the status games, especially in SF, where ML researchers are a higher tier than ML engineers, which are a higher tier than software engineers. I think that is completely messed up. My hope is that we can create a new type of role: a software researcher, someone who actually pushes the boundaries of software.” @CompleteSkeptic Love to hear your thoughts @vboykis @eugeneyan @sh_reya @jeremyphoward

0:20